更新 .gitea/workflows/prod_build_manual.yaml

# Conflicts:
#	app/service/design_fast/utils/synthesis_item.py
#	app/service/prompt_generation/chatgpt_for_translation.py

.dockerignore

@@ -1,2 +1,6 @@
 seg_cache
-test
+test
+.venv
+__pycache__/
+*.pyc
+.git/

.gitea/workflows/develop_build_manual.yaml

@@ -7,7 +7,7 @@ jobs:
     runs-on: ubuntu-latest
 
     env:
-      REMOTE_DEPLOY_PATH: /workspace/Trinity/Fastapi_AiDA_Trinity_Dev
+      REMOTE_DEPLOY_PATH: /workspace/AiDA_Workspace/Python_Server_Workspace/Dev
 
     steps:
       - name: 1.检出代码
@@ -35,6 +35,4 @@ jobs:
             cd ${{ env.REMOTE_DEPLOY_PATH }}
             
             docker-compose down 2>&1
-            docker-compose up -d --build --remove-orphans 2>&1
+            docker-compose up -d  2>&1
-            
-            docker image prune -f 2>&1

.gitea/workflows/develop_build_scheduled.yaml

@@ -1,15 +1,15 @@
 name: 定时 AiDA python develop 分支构建部署
 on:
   # 使用 schedule 触发器，遵循标准的 Cron 格式 (分钟 小时-8 日期 月份 星期)
-  schedule:
+  # schedule:
-    - cron: '30 9 * * *'
+  #   - cron: '30 9 * * *'
 
 jobs:
   scheduled_deploy:
     runs-on: ubuntu-latest
 
     env:
-      REMOTE_DEPLOY_PATH: /workspace/Trinity/Fastapi_AiDA_Trinity_Dev
+      REMOTE_DEPLOY_PATH: /workspace/AiDA_Workspace/Python_Server_Workspace/Dev
 
     steps:
       - name: 1.检出代码

.gitea/workflows/ltx_develop_build_manual.yaml

@@ -1,23 +1,19 @@
-name: git commit AiDA python develop 分支构建部署
+name: 手动 AiDA python develop 分支构建部署
 on:
   workflow_dispatch:
-  push:
-    branches:
-      - develop
 
 jobs:
   scheduled_deploy:
     runs-on: ubuntu-latest
-    if: "contains(github.event.head_commit.message, '[run build]')"
 
     env:
-      REMOTE_DEPLOY_PATH: /workspace/Trinity/Fastapi_AiDA_Trinity_Dev
+      REMOTE_DEPLOY_PATH: /workspace/AiDA_Workspace/Python_Server_Workspace/Dev
 
     steps:
       - name: 1.检出代码
         uses: actions/checkout@v4
         with:
-          ref: 'develop'
+          ref: 'dev-ltx'
 
       - name: 2.复制文件到服务器
         uses: appleboy/scp-action@v0.1.7
@@ -28,7 +24,7 @@ jobs:
           source: "."
           target: ${{ env.REMOTE_DEPLOY_PATH }}
       
-      - name: Restart Docker containers
+      - name: 3.重启docker-compose
         uses: appleboy/ssh-action@v0.1.10
         with:
           host: ${{ secrets.SERVER_HOST }}

.gitea/workflows/prod_build_manual.yaml

@@ -0,0 +1,40 @@
+name: 定时 AiDA python prod 分支构建部署
+on:
+  workflow_dispatch:
+
+jobs:
+  scheduled_deploy:
+    runs-on: ubuntu-latest
+
+    env:
+      REMOTE_DEPLOY_PATH: /workspace/AiDA_Workspace/Python_Server_Workspace/AiDA_Prod
+
+    steps:
+      - name: 1.检出代码
+        uses: actions/checkout@v4
+        with:
+          ref: 'master'
+
+      - name: 2.复制文件到服务器
+        uses: appleboy/scp-action@v0.1.7
+        with:
+          host: ${{ secrets.SERVER_HOST }}
+          username: ${{ secrets.SERVER_USER }}
+          password: ${{ secrets.SERVER_PASSWORD }}
+          source: "."
+          target: ${{ env.REMOTE_DEPLOY_PATH }}
+      
+      - name: Restart Docker containers
+        uses: appleboy/ssh-action@v0.1.10
+        with:
+          host: ${{ secrets.SERVER_HOST }}
+          username: ${{ secrets.SERVER_USER }}
+          password: ${{ secrets.SERVER_PASSWORD }}
+          script: |
+            # 进入项目目录
+            cd ${{ env.REMOTE_DEPLOY_PATH }}
+            
+            docker-compose down 2>&1
+            docker-compose up -d 2>&1
+            
+            docker image prune -f 2>&1

.gitea/workflows/prod_build_scheduled.yaml

@@ -0,0 +1,42 @@
+name: 定时 AiDA python prod 分支构建部署
+on:
+  # 使用 schedule 触发器，遵循标准的 Cron 格式 (分钟 小时-8 日期 月份 星期)
+  schedule:
+    - cron: '07 13 23 1 *'
+
+jobs:
+  scheduled_deploy:
+    runs-on: ubuntu-latest
+
+    env:
+      REMOTE_DEPLOY_PATH: /workspace/AiDA_Workspace/Python_Server_Workspace/Prod
+
+    steps:
+      - name: 1.检出代码
+        uses: actions/checkout@v4
+        with:
+          ref: 'master'
+
+      - name: 2.复制文件到服务器
+        uses: appleboy/scp-action@v0.1.7
+        with:
+          host: ${{ secrets.SERVER_HOST }}
+          username: ${{ secrets.SERVER_USER }}
+          password: ${{ secrets.SERVER_PASSWORD }}
+          source: "."
+          target: ${{ env.REMOTE_DEPLOY_PATH }}
+      
+      - name: Restart Docker containers
+        uses: appleboy/ssh-action@v0.1.10
+        with:
+          host: ${{ secrets.SERVER_HOST }}
+          username: ${{ secrets.SERVER_USER }}
+          password: ${{ secrets.SERVER_PASSWORD }}
+          script: |
+            # 进入项目目录
+            cd ${{ env.REMOTE_DEPLOY_PATH }}
+            
+            docker-compose down 2>&1
+            docker-compose up -d 2>&1
+            
+            docker image prune -f 2>&1

README.md

@@ -20,7 +20,6 @@
          $ conda activate trinity_client_aida 
          $ pip install -r requirements.txt
          $ conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia -y
-         $ pip install mmcv==1.4.2 -f https://download.openmmlab.com/mmcv/dist/cu117/torch1.13/index.html
 
 
 1. 启动服务器

app/api/api_design.py

@@ -1,9 +1,11 @@
 import json
 import logging
 
+import requests
 from fastapi import APIRouter, HTTPException, BackgroundTasks
 
-from app.schemas.design import DesignModel, ModelProgressModel, DesignStreamModel
+from app.core.config import settings
+from app.schemas.design import DesignModel, ModelProgressModel, DesignStreamModel, SAMRequestModel
 from app.schemas.response_template import ResponseModel
 from app.service.design_fast.design_generate import design_generate, design_generate_v2
 from app.service.design_fast.model_process_service import model_transpose
@@ -15,177 +17,141 @@ logger = logging.getLogger()
 @router.post("/design")
 def design(request_data: DesignModel):
     """
-            objects.items.transparent:
+    - **objects.items.transparent**:
-                "transparent":{
+    ```json
-                    "mask_url":"test/transparent_test/transparent_mask.png",
+        "transparent":{
-                    "scale":0.1
+            "mask_url":"test/transparent_test/transparent_mask.png",
-                },
+            "scale":0.1
-                mask_url 为空"" ->  单件衣服透明
+        },
-                mask_url 非空"mask_url" ->  区域透明
+    ```
+    - **mask_url** 为空"" ->  单件衣服透明
+    - **mask_url** 非空"mask_url" ->  区域透明
+    - **transpose**  镜像模式 ,:"top_bottom"或"left_right"
+    - **rotate** 45,
 
-            创建一个具有以下参数的请求体:
+    - ** design 参数变更:
-            示例参数：
+            design detail 请求参数中 basic -> preview_submit 替换为design_type  可选参数 default ,merge   (移除preview和submit)
-            {
+         design_type 参数说明:
-                "objects": [
+            defuault模式下 请求参数不变
-                    {
+            merge模式下 items -> 每个item需要新增 merge_image_path , merge_image_path为前端处理 print  color等操作后的单件结果图
-                        "basic": {
+
-                            "body_point_test": {
+    **
-                                "waistband_right": [
+
-                                    203,
+    - 创建一个具有以下参数的请求体:
-                                    249
+    示例参数：
-                                ],
+    ```json
-                                "hand_point_right": [
+        {
-                                    229,
+            "objects": [
-                                    343
+                {
-                                ],
+                    "basic": {
-                                "waistband_left": [
+                        "body_point_test": {
-                                    119,
+                            "waistband_right": [
-                                    248
+                                203,
-                                ],
+                                249
-                                "hand_point_left": [
+                            ],
-                                    97,
+                            "hand_point_right": [
-                                    343
+                                229,
-                                ],
+                                343
-                                "shoulder_left": [
+                            ],
-                                    108,
+                            "waistband_left": [
-                                    107
+                                119,
-                                ],
+                                248
-                                "shoulder_right": [
+                            ],
-                                    212,
+                            "hand_point_left": [
-                                    107
+                                97,
-                                ]
+                                343
-                            },
+                            ],
-                            "layer_order": true,
+                            "shoulder_left": [
-                            "preview_submit": "submit",
+                                108,
-                            "scale_bag": 0.7,
+                                107
-                            "scale_earrings": 0.16,
+                            ],
-                            "self_template": true,
+                            "shoulder_right": [
-                            "single_overall": "overall",
+                                212,
-                            "switch_category": ""
+                                107
+                            ]
                         },
-                        "items": [
+                        "layer_order": true,
-                            {
+                        "design_type": "preview",
-                                "businessId": 2377945,
+                        "scale_bag": 0.7,
-                                "color": "209 196 171",
+                        "scale_earrings": 0.16,
-                                "image_id": 189410,
+                        "self_template": true,
-                                "offset": [
+                        "single_overall": "overall",
-                                    0,
+                        "switch_category": ""
-                                    0
+                    },
-                                ],
+                    "items": [
-                                "path": "aida-collection-element/89/Sketchboard/53d38bd5-f77b-4034-ada2-45f1e2ebe00c.png",
+                        {
-                                "print": {
+                            "businessId": 2115382,
-                                    "element": {
+                            "color": "",
-                                        "element_angle_list": [],
+                            "image_id": 61686,
-                                        "element_path_list": [],
+                            "offset": [
-                                        "element_scale_list": [],
+                                0,
-                                        "location": []
+                                0
-                                    },
+                            ],
-                                    "overall": {
+                            "path": "aida-sys-image/images/female/dress/0628000564.jpg",
-                                        "location": [],
+                            "transpose": [
-                                        "print_angle_list": [],
+                                1,
-                                        "print_path_list": [],
+                                1
-                                        "print_scale_list": []
+                            ],
-                                    },
+                            "rotate": 45,
-                                    "single": {
+                            "print": {
-                                        "location": [],
+                                "element": {
-                                        "print_angle_list": [],
+                                    "element_angle_list": [],
-                                        "print_path_list": [],
+                                    "element_path_list": [],
-                                        "print_scale_list": []
+                                    "element_scale_list": [],
-                                    }
+                                    "location": []
                                 },
-                                "priority": 12,
+                                "overall": {
-                                "resize_scale": [
+                                    "location": [
-                                    1.0,
+                                        [
-                                    1.0
+                                            53.0,
-                                ],
+                                            118.5
-                                "seg_mask_url": "aida-clothing/mask/mask_8e96ddb0-e466-11f0-8de2-0242ac130002.png",
+                                        ]
-                                "type": "Outwear"
+                                    ],
-                            },
+                                    "print_angle_list": [
-                            {
+                                        0.0
-                                "businessId": 2377946,
+                                    ],
-                                "color": "122 152 139",
+                                    "print_path_list": [
-                                "image_id": 81868,
+                                        "aida-users/89/print/02d57aa8-f342-4e1d-b02c-b278f94dcfe6-3-89.png"
-                                "offset": [
+                                    ],
-                                    0,
+                                    "print_scale_list": [
-                                    0
+                                        [
-                                ],
+                                            0.5,
-                                "path": "aida-sys-image/images/female/blouse/0825001443.jpg",
+                                            0.5
-                                "print": {
+                                        ]
-                                    "element": {
+                                    ],
-                                        "element_angle_list": [],
+                                    "gap": [
-                                        "element_path_list": [],
+                                        [
-                                        "element_scale_list": [],
+                                            10,
-                                        "location": []
+                                            10
-                                    },
+                                        ]
-                                    "overall": {
+                                    ]
-                                        "location": [],
-                                        "print_angle_list": [],
-                                        "print_path_list": [],
-                                        "print_scale_list": []
-                                    },
-                                    "single": {
-                                        "location": [],
-                                        "print_angle_list": [],
-                                        "print_path_list": [],
-                                        "print_scale_list": []
-                                    }
                                 },
-                                "priority": 11,
+                                "single": {
-                                "resize_scale": [
+                                    "location": [],
-                                    1.0,
+                                    "print_angle_list": [],
-                                    1.0
+                                    "print_path_list": [],
-                                ],
+                                    "print_scale_list": []
-                                "seg_mask_url": "aida-clothing/mask/mask_8f0fab78-e466-11f0-8de2-0242ac130002.png",
+                                }
-                                "type": "Blouse"
                             },
-                            {
+                            "priority": 10,
-                                "businessId": 2377947,
+                            "resize_scale": [
-                                "color": "111 78 63",
+                                1.0,
-                                "gradient": "aida-gradient/517c3a4d-aed7-4423-aa99-7b60d3577df1.png",
+                                1.0
-                                "image_id": 116494,
+                            ],
-                                "offset": [
+                            "seg_mask_url": "aida-clothing/mask/mask_9698b428-eb93-11f0-9327-0242c0a80003.png",
-                                    0,
+                            "type": "Dress"
-                                    0
+                        },
-                                ],
+                        {
-                                "path": "aida-sys-image/images/female/skirt/0825000219.jpg",
+                            "body_path": "aida-sys-image/models/female/2e4815b9-1191-419d-94ed-5771239ca4a5.png",
-                                "print": {
+                            "image_id": 67277,
-                                    "element": {
+                            "type": "Body"
-                                        "element_angle_list": [],
+                        }
-                                        "element_path_list": [],
+                    ]
-                                        "element_scale_list": [],
+                }
-                                        "location": []
+            ],
-                                    },
+            "process_id": "89"
-                                    "overall": {
+        }
-                                        "location": [],
+    ```
-                                        "print_angle_list": [],
-                                        "print_path_list": [],
-                                        "print_scale_list": []
-                                    },
-                                    "single": {
-                                        "location": [],
-                                        "print_angle_list": [],
-                                        "print_path_list": [],
-                                        "print_scale_list": []
-                                    }
-                                },
-                                "priority": 10,
-                                "resize_scale": [
-                                    1.0,
-                                    1.0
-                                ],
-                                "seg_mask_url": "aida-clothing/mask/mask_8f6191fe-e466-11f0-8de2-0242ac130002.png",
-                                "type": "Skirt"
-                            },
-                            {
-                                "body_path": "aida-sys-image/models/female/2e4815b9-1191-419d-94ed-5771239ca4a5.png",
-                                "image_id": 67277,
-                                "type": "Body"
-                            }
-                        ]
-                    }
-                ],
-                "process_id": "89"
-            }
         """
     # logger.info(f"design request item is : @@@@@@:{json.dumps(request_data.dict(),indent=4)}")
     # data = generate(request_data=request_data)
@@ -421,6 +387,55 @@ async def design_v2(request_data: DesignStreamModel, background_tasks: Backgroun
     return ResponseModel()
 
 
+@router.post("/seg_anything")
+async def seg_anything(request_data: SAMRequestModel):
+    """
+    **Segment Anything 交互式分割接口**
+
+    通过传入图片路径和点击的点坐标，返回分割后的掩码数据。
+
+    ### 参数说明:
+    - **bucket**: minio bucket name
+    - **object_name**: minio object name
+    - **image_path**: 图片在服务器或云端的相对路径。
+    - **type**: 推理类型
+    - **box**: 框选矩形点位信息
+    - **points**: 交互点的坐标列表。每个点为 [x, y] 像素格式。
+    - **labels**: 坐标点的属性标签，必须与 points 长度一致：
+        - 1: **前景点** (代表想要分割出的区域)
+        - 0: **背景点** (代表想要排除的区域)
+
+    ### 请求体示例:
+    ```json
+    point
+    {
+        "bucket": "test",
+        "object_name": "7068-400a-ac94-c01647fa5f6f.png",
+        "image_path": "aida-users/89/sketch/4e8fe37d-7068-400a-ac94-c01647fa5f6f.png",
+        "type":"point",
+        "points": [[310, 403], [493, 375], [261, 266], [404, 484]],
+        "labels": [1, 1, 0, 1]
+    }
+
+    box
+    {
+        "bucket": "test",
+        "object_name": "7068-400a-ac94-c01647fa5f6f.png",
+        "image_path": "aida-users/89/sketch/4e8fe37d-7068-400a-ac94-c01647fa5f6f.png",
+        "type":"box",
+        "box": [350, 286, 544, 520]
+    }
+    ```
+    """
+    try:
+        logger.info(f"seg_anything request item is : @@@@@@:{json.dumps(request_data.dict(), indent=4)}")
+        data = requests.post(f"http://{settings.B_4_X_4090_SERVICE_HOST}:10075/predict", json=request_data.dict())
+        logger.info(f"seg_anything response @@@@@@:{json.dumps(json.loads(data.content), indent=4)}")
+        return ResponseModel(data=json.loads(data.content))
+    except Exception as e:
+        logger.warning(f"seg_anything Run Exception @@@@@@:{e}")
+
+
 # @router.post('/get_progress')
 # def get_progress(request_data: DesignProgressModel):
 #     """

app/api/api_generate_image.py

@@ -1,9 +1,12 @@
 import json
 import logging
 
+import httpx
+import requests
 from fastapi import APIRouter, BackgroundTasks, HTTPException
 
-from app.schemas.generate_image import GenerateImageModel, GenerateProductImageModel, GenerateSingleLogoImageModel, GenerateRelightImageModel, GenerateMultiViewModel, BatchGenerateProductImageModel, BatchGenerateRelightImageModel, AgentTollGenerateImageModel
+from app.core.config import settings
+from app.schemas.generate_image import GenerateImageModel, GenerateProductImageModel, GenerateSingleLogoImageModel, GenerateRelightImageModel, GenerateMultiViewModel, BatchGenerateProductImageModel, BatchGenerateRelightImageModel, AgentTollGenerateImageModel, Flux2ToProductImgModel, GenerateSloganImageModel, GenerateImageFlux2KleinModel
 from app.schemas.pose_transform import BatchPoseTransformModel
 from app.schemas.response_template import ResponseModel
 from app.service.generate_batch_image.service import start_product_batch_generate, start_relight_batch_generate, start_pose_transform_batch_generate
@@ -20,6 +23,61 @@ logger = logging.getLogger()
 '''generate image'''
 
 
+# flux2 klein
+@router.post("/generate_image_flux2_klein")
+async def generate_image_flux2_klein(request_item: GenerateImageFlux2KleinModel):
+    """
+        创建一个具有以下参数的请求体:
+        - **bucket_name**: OSS桶名 (必填)
+        - **object_name**: OSS对象名（文件路径）(必填)
+
+        - **width**: 图片宽度，默认1024像素 (非必填,1024)
+        - **height**: 图片高度，默认1024像素 (非必填,默认1024)
+        - **prompt**: 文本提示词，用于模型推理等场景 (非必填,默认"")
+        - **steps**: 推理步数，控制模型生成过程的迭代次数 (非必填,默认4)
+        - **guidance**: 引导系数，调节提示词对生成结果的影响程度 (非必填,默认 4.0 )
+
+        ### 示例参数：
+        ```
+            {
+                "bucket_name": "aida-users",
+                "object_name": "89/moodboard/5fdc698c-cb9b-4b36-afa9ce4-1-89.png",
+                "prompt": "a single item of sketch of dress, 4k, white background"
+            }
+        ```
+        ### 输出示例:
+        ```
+            {
+                "code": 200,
+                "msg": "OK!",
+                "data": {
+                    "output_path": "aida-users/89/moodboard/5fdc698c-cb9b-4b36-afa9ce4-1-89.png"
+                }
+            }
+        ```
+    """
+    try:
+        logger.info(f"generate_image_flux2_gen_img request: {json.dumps(request_item.model_dump(), indent=4)}")
+        async with httpx.AsyncClient(timeout=120) as client:
+            resp = await client.post(
+                f"http://{settings.FLUX2_GEN_IMG_MODEL_URL}/predict",
+                json=request_item.model_dump(),
+            )
+        if resp.status_code == 200:
+            result = resp.json()
+            logger.info(f"flux2_gen_img response: {json.dumps(result, indent=4)}")
+            return ResponseModel(data=result)
+        else:
+            error = resp.json()
+            logger.info(f"flux2_gen_img response: {json.dumps(error, indent=4)}")
+            return ResponseModel(data=error, msg="ERROR!", code=500)
+
+    except Exception as e:
+        logger.warning(f"generate_image_flux2_gen_img Run Exception @@@@@@:{e}")
+        raise HTTPException(status_code=404, detail=str(e))
+
+
+# sdxl
 @router.post("/generate_image")
 def generate_image(request_item: GenerateImageModel, background_tasks: BackgroundTasks):
     """
@@ -154,6 +212,62 @@ def generate_single_logo_image(tasks_id: str):
     return ResponseModel(data=data['data'])
 
 
+"""slogan """
+
+
+@router.post("/generate_slogan")
+async def generate_slogan(request_data: GenerateSloganImageModel):
+    """
+        ### 请求体示例:
+        ```json
+        {
+          "num_point": 16,
+          "image_url": "aida-slogan/6886785f-0aac-4052-b6fd-7ae20a841d8d.png",
+          "prompt": "123",
+          "tasks_id": "string-89"
+        }
+        ```
+    """
+    try:
+        logger.info(f"generate_slogan request item is : @@@@@@:{json.dumps(request_data.dict(), indent=4)}")
+        data = requests.post(f"http://{settings.A6000_SERVICE_HOST}:10020/api/slogan", json=request_data.dict())
+        logger.info(f"generate_slogan response @@@@@@:{json.dumps(json.loads(data.content), indent=4)}")
+        return ResponseModel(data=json.loads(data.content))
+    except Exception as e:
+        logger.warning(f"generate_slogan Run Exception @@@@@@:{e}")
+
+
+"""product image flux2.0"""
+
+# @router.post("/img_to_product")
+# async def img_to_product(request_data: Flux2ToProductImgModel):
+#     """
+#         创建一个具有以下参数的请求体:
+#         - **tasks_id**: 任务id 用于取消生成任务和获取生成结果
+#         - **prompt**: 想要生成图片的描述词
+#         - **image_path**: 被生成图片的S3或minio url地址
+#         - **infer_step**: 推理步数
+#
+#         ### 请求体示例:
+#         ```json
+#         point
+#         {
+#             "prompt": "Create realistic studio photo with real people model standing and wearing this garment, in white studio, Keep original model if present, or generate appropriate model, Standing pose, facing camera.",
+#             "image_path":"aida-results/result_38151e0a-f83b-11f0-89f6-0242ac130002.png",
+#             "infer_step":4,
+#             "tasks_id":"123456-123"
+#         }
+#         ```
+#     """
+#     try:
+#         logger.info(f"img_to_product request item is : @@@@@@:{json.dumps(request_data.dict(), indent=4)}")
+#         data = requests.post(f"http://{settings.A6000_SERVICE_HOST}:10090/api/v1/to_product", json=request_data.dict())
+#         logger.info(f"img_to_product response @@@@@@:{json.dumps(json.loads(data.content), indent=4)}")
+#         return ResponseModel(data=json.loads(data.content))
+#     except Exception as e:
+#         logger.warning(f"img_to_product Run Exception @@@@@@:{e}")
+
+
 '''product image'''
 
 
@@ -178,7 +292,7 @@ def generate_product_image(request_item: GenerateProductImageModel, background_t
         }
     """
     try:
-        logger.info(f"generate_product_image request item is : @@@@@@:{json.dumps(request_item.dict(),indent=4)}")
+        logger.info(f"generate_product_image request item is : @@@@@@:{json.dumps(request_item.dict(), indent=4)}")
         service = GenerateProductImage(request_item)
         background_tasks.add_task(service.get_result)
     except Exception as e:

app/api/api_recommendation.py

@@ -137,10 +137,13 @@ router = APIRouter()
 #         logger.error(f"推荐失败: {str(e)}", exc_info=True)
 #         raise HTTPException(status_code=500, detail=str(e))
 
-# @router.on_event("startup")
+@router.on_event("startup")
 async def startup_event():
     """启动时初始化增量监听任务"""
     try:
+        # 屏蔽 apscheduler 的 INFO 日志
+        logging.getLogger("apscheduler").setLevel(logging.WARNING)
+
         # 确保 Milvus 集合已创建（若已存在则直接返回）
         try:
             create_collection()
@@ -172,4 +175,32 @@ async def recommend(
         return [path]
     except Exception as e:
         logger.error("新版推荐接口失败 [user=%s, category=%s]: %s", user_id, category, e, exc_info=True)
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+@router.get("/redis/user_pref")
+async def get_all_user_preferences():
+    """
+    获取所有以 user_pref 为前缀的 Redis key 信息
+    """
+    try:
+        from app.service.utils.redis_utils import Redis
+        from app.service.recommendation_system.config import REDIS_KEY_USER_PREF_PREFIX
+        
+        # 扫描所有匹配 user_pref:* 的 key
+        pattern = f"{REDIS_KEY_USER_PREF_PREFIX}:*"
+        keys = Redis.scan_keys(pattern)
+        
+        # 直接返回所有 key 和原始 value
+        result = {}
+        for key in keys:
+            # 读取对应的值
+            value = Redis.read(key)
+            if value:
+                result[key] = value
+        
+        return result
+        
+    except Exception as e:
+        logger.error("获取用户偏好数据失败: %s", e, exc_info=True)
         raise HTTPException(status_code=500, detail=str(e))

app/api/api_route.py

@@ -7,6 +7,7 @@ from app.api import api_design_pre_processing
 from app.api import api_generate_image
 from app.api import api_mannequins_edit
 from app.api import api_pose_transform
+from app.api import api_precompute
 from app.api import api_prompt_generation
 from app.api import api_recommendation
 from app.api import api_test
@@ -21,6 +22,7 @@ router.include_router(api_prompt_generation.router, tags=['prompt_generation'],
 router.include_router(api_design_pre_processing.router, tags=['design_pre_processing'], prefix="/api")
 router.include_router(api_brand_dna.router, tags=['api_brand_dna'], prefix="/api")
 router.include_router(api_recommendation.router, tags=['api_recommendation'], prefix="/api")
+router.include_router(api_precompute.router, tags=['api_precompute'], prefix="/api")
 router.include_router(api_mannequins_edit.router, tags=['api_mannequins_edit'], prefix="/api")
 router.include_router(api_pose_transform.router, tags=['api_pose_transform'], prefix="/api")
 router.include_router(api_clothing_seg.router, tags=['api_clothing_seg'], prefix="/api")

app/core/config.backup.py

@@ -1,235 +0,0 @@
-import os
-
-import pika
-from dotenv import load_dotenv
-from pydantic import BaseSettings
-
-BASE_DIR = os.path.abspath(os.path.join(os.path.dirname(__file__), '../../'))
-load_dotenv(os.path.join(BASE_DIR, '.env'))
-
-
-class Settings(BaseSettings):
-    PROJECT_NAME: str = 'FASTAPI BASE'
-    SECRET_KEY: str = ''
-    API_PREFIX: str = ''
-    BACKEND_CORS_ORIGINS: list[str] = ['*']
-    DATABASE_URL: str = ''
-    ACCESS_TOKEN_EXPIRE_SECONDS: int = 60 * 60 * 24 * 7  # Token expired after 7 days
-    SECURITY_ALGORITHM: str = 'HS256'
-    LOGGING_CONFIG_FILE: str = os.path.join(BASE_DIR, 'logging_env.py')
-
-
-OSS = "minio"
-DEBUG = False
-if DEBUG:
-    LOGS_PATH = "logs/"
-    CATEGORY_PATH = "service/attribute/config/descriptor/category/category_dis.csv"
-    SEG_CACHE_PATH = "../seg_cache/"
-    POSE_TRANSFORM_VIDEO_PATH = "../pose_transform_video/"
-    RECOMMEND_PATH_PREFIX = "service/recommend/"
-    CHROMADB_PATH = "./chromadb/"
-else:
-    LOGS_PATH = "app/logs/"
-    CATEGORY_PATH = "app/service/attribute/config/descriptor/category/category_dis.csv"
-    SEG_CACHE_PATH = "/seg_cache/"
-    POSE_TRANSFORM_VIDEO_PATH = "/pose_transform_video/"
-    RECOMMEND_PATH_PREFIX = "app/service/recommend/"
-    CHROMADB_PATH = "/chromadb/"
-
-# RABBITMQ_ENV = ""  # 生产环境
-RABBITMQ_ENV = os.getenv("RABBITMQ_ENV", "-dev")
-# RABBITMQ_ENV = "-local"  # 本地测试环境
-
-if RABBITMQ_ENV == "-dev":
-    JAVA_STREAM_API_URL = f"https://develop.api.aida.com.hk/api/third/party/receiveDesignResults"
-elif RABBITMQ_ENV == "-prod":
-    JAVA_STREAM_API_URL = f"https://api.aida.com.hk/api/third/party/receiveDesignResults"
-
-settings = Settings()
-
-# minio 配置
-MINIO_URL = "www.minio-api.aida.com.hk"
-MINIO_ACCESS = 'vXKFLSJkYeEq2DrSZvkB'
-MINIO_SECRET = 'uKTZT3x7C43WvPN9QTc99DiRkwddWZrG9Uh3JVlR'
-MINIO_SECURE = True
-
-# S3 配置
-S3_ACCESS_KEY = "AKIAVD3OJIMF6UJFLSHZ"
-S3_AWS_SECRET_ACCESS_KEY = "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"
-S3_REGION_NAME = "ap-east-1"
-
-# redis 配置
-REDIS_HOST = "10.1.1.240"
-REDIS_PORT = "6379"
-REDIS_DB = "2"
-
-# rabbitmq config
-RABBITMQ_PARAMS = {
-    "host": "18.167.251.121",
-    "port": 5672,
-    "credentials": pika.credentials.PlainCredentials(username='rabbit', password='123456'),
-    "virtual_host": "/"
-}
-
-# milvus 配置
-MILVUS_URL = "http://10.1.1.240:19530"
-MILVUS_TOKEN = "root:Milvus"
-MILVUS_ALIAS = "default"
-MILVUS_TABLE_KEYPOINT = "keypoint_cache_2"
-MILVUS_TABLE_SEG = "seg_cache"
-
-# Mysql 配置
-DB_HOST = '18.167.251.121'  # 数据库主机地址
-# DB_PORT = int( 33006)
-DB_PORT = 33008  # 数据库端口
-DB_USERNAME = 'aida_con_python'  # 数据库用户名
-DB_PASSWORD = '123456'  # 数据库密码
-DB_NAME = 'aida'  # 数据库库名
-
-# openai
-os.environ['SERPAPI_API_KEY'] = "a793513017b0718db7966207c31703d280d12435c982f1e67bbcbffa52e7632c"
-OPENAI_STREAM = True
-BUFFER_THRESHOLD = 6  # must be even number
-SINGLE_TOKEN_THRESHOLD = 200
-TOKEN_THRESHOLD = 600
-OPENAI_TEMPERATURE = 0
-
-# OPENAI_API_KEY =  "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"
-OPENAI_API_KEY = "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"
-OPENAI_MODEL = "gpt-3.5-turbo-0613"
-OPENAI_MODEL_LIST = {"gpt-3.5-turbo-0613",
-                     "gpt-3.5-turbo-16k-0613",
-                     "gpt-4-0314",
-                     "gpt-4-32k-0314",
-                     "gpt-4-0613",
-                     "gpt-4-32k-0613", }
-
-# SR service config
-SR_MODEL_NAME = "super_resolution"
-SR_TRITON_URL = "10.1.1.240:10031"
-SR_MINIO_BUCKET = "aida-users"
-SR_RABBITMQ_QUEUES = f"SuperResolution{RABBITMQ_ENV}"
-
-# GenerateImage service config
-FAST_GI_MODEL_URL = '10.1.1.243:10011'
-FAST_GI_MODEL_NAME = 'stable_diffusion_xl'
-
-GI_MODEL_URL = '10.1.1.240:10061'
-GI_MODEL_NAME = 'flux'
-
-GMV_MODEL_URL = '10.1.1.243:10081'
-GMV_MODEL_NAME = 'multi_view'
-
-GMV_RABBITMQ_QUEUES = f"GenerateMultiView{RABBITMQ_ENV}"
-
-GI_MINIO_BUCKET = "aida-users"
-GI_RABBITMQ_QUEUES = f"GenerateImage{RABBITMQ_ENV}"
-GI_SYS_IMAGE_URL = "aida-sys-image/generate_image/white_image.jpg"
-
-# SLOGAN service config
-SLOGAN_RABBITMQ_QUEUES = f"Slogan{RABBITMQ_ENV}"
-
-# Generate Single Logo service config
-GSL_MODEL_URL = '10.1.1.243:10041'
-GSL_MINIO_BUCKET = "aida-users"
-GSL_MODEL_NAME = 'stable_diffusion_xl_transparent'
-GEN_SINGLE_LOGO_RABBITMQ_QUEUES = f"GenSingleLogo{RABBITMQ_ENV}"
-
-# Generate Product service config
-# GPI_RABBITMQ_QUEUES = os.getenv("GEN_PRODUCT_IMAGE_RABBITMQ_QUEUES", f"ToProductImage{RABBITMQ_ENV}")
-# GPI_MODEL_NAME_OVERALL = 'sdxl_ensemble_all'
-# GPI_MODEL_URL = '10.1.1.243:10051'
-
-# Generate Product service config 旧版product img 模型
-GPI_RABBITMQ_QUEUES = f"ToProductImage{RABBITMQ_ENV}"
-BATCH_GPI_RABBITMQ_QUEUES = f"BatchToProductImage{RABBITMQ_ENV}"
-GPI_MODEL_NAME_OVERALL = 'diffusion_ensemble_all'
-GPI_MODEL_NAME_SINGLE = 'stable_diffusion_1_5_cnet'
-GPI_MODEL_URL = '10.1.1.243:10051'
-
-# Generate Single Logo service config
-GRI_RABBITMQ_QUEUES = f"Relight{RABBITMQ_ENV}"
-BATCH_GRI_RABBITMQ_QUEUES = f"BatchRelight{RABBITMQ_ENV}"
-GRI_MODEL_NAME_OVERALL = 'diffusion_relight_ensemble'
-GRI_MODEL_NAME_SINGLE = 'stable_diffusion_1_5_relight'
-GRI_MODEL_URL = '10.1.1.240:10051'
-
-# Pose Transform service config
-
-PS_RABBITMQ_QUEUES = f"PoseTransform{RABBITMQ_ENV}"
-BATCH_PS_RABBITMQ_QUEUES = f"BatchPoseTransform{RABBITMQ_ENV}"
-PT_MODEL_URL = '10.1.1.243:10061'
-
-# SEG service config
-SEGMENTATION = {
-    "new_model_name": "seg_knet",
-    "name": "seg_ocrnet_hr18",
-    "input": "seg_input__0",
-    "output": "seg_output__0",
-}
-# ollama config
-OLLAMA_URL = "http://10.1.1.240:11434/api/embeddings"
-
-# design batch
-BATCH_DESIGN_RABBITMQ_QUEUES = f"DesignBatch{RABBITMQ_ENV}"
-
-# DESIGN config
-DESIGN_MODEL_URL = '10.1.1.240:10000'
-AIDA_CLOTHING = "aida-clothing"
-KEYPOINT_RESULT_TABLE_FIELD_SET = ('neckline_left', 'neckline_right', 'shoulder_left', 'shoulder_right', 'armpit_left', 'armpit_right',
-                                   'cuff_left_in', 'cuff_left_out', 'cuff_right_in', 'cuff_right_out', 'waistband_left', 'waistband_right')
-
-# DESIGN 预处理
-IF_DEBUG_SHOW = False
-
-# 优先级
-PRIORITY_DICT = {
-    'earring_front': 99,
-    'bag_front': 98,
-    'hairstyle_front': 97,
-    'outwear_front': 20,
-    'tops_front': 19,
-    'dress_front': 18,
-    'blouse_front': 17,
-    'skirt_front': 16,
-    'trousers_front': 15,
-    'bottoms_front': 14,
-    'shoes_right': 1,
-    'shoes_left': 1,
-    'body': 0,
-    'bottoms_back': -14,
-    'trousers_back': -15,
-    'skirt_back': -16,
-    'blouse_back': -17,
-    'dress_back': -18,
-    'tops_back': -19,
-    'outwear_back': -20,
-    'hairstyle_back': -97,
-    'bag_back': -98,
-    'earring_back': -99,
-}
-
-QWEN_API_KEY = "sk-f31c29e61ac2498ba5e307aaa6dc10e0"
-
-DB_CONFIG = {
-    "host": "18.167.251.121",
-    "port": 3306,
-    "user": "root",
-    "password": "QWa998345",
-    "database": "aida",
-    "charset": "utf8mb4"
-}
-
-TABLE_CATEGORIES = {
-    "female_dress": "female/dress",
-    "female_outwear": "female/outwear",
-    "female_trousers": "female/trousers",
-    "female_skirt": "female/skirt",
-    "female_blouse": "female/blouse",
-    "male_tops": "male/tops",
-    "male_bottoms": "male/bottoms",
-    "male_outwear": "male/outwear"
-}
-
-# --- ComfyUI 配置信息 ---
-COMFYUI_SERVER_ADDRESS = "10.1.2.227:8080"  # 替换为您的 ComfyUI 服务器地址

app/core/config.py

@@ -36,7 +36,7 @@ class Settings(BaseSettings):
 
     # --- mysql 配置信息 ---
     MYSQL_HOST: str = Field(default='', description="")
-    MYSQL_PORT: str = Field(default='', description="")
+    MYSQL_PORT: int = Field(default=3306, description="")
     MYSQL_USER: str = Field(default='', description="")
     MYSQL_PASSWORD: str = Field(default='', description="")
     MYSQL_DB: str = Field(default='', description="")
@@ -64,11 +64,19 @@ class Settings(BaseSettings):
     # --- Design Callback Java 接口 ---
     JAVA_STREAM_API_URL: str = Field(default='', description="")
 
+    # ---  flux2 klein model url ---
+    FLUX2_GEN_IMG_MODEL_URL: str = Field(default='', description="")
+
+    # --- 服务器IP ---
+    A6000_SERVICE_HOST: str = Field(default='', description="")
+    B_4_X_4090_SERVICE_HOST: str = Field(default='', description="")
+
     # --- 其他配置信息 以下均为Docker容器内配置---
     LOGS_PATH: str = Field(default="/logs/", description="")
     CATEGORY_PATH: str = Field(default="/app/service/attribute/config/descriptor/category/category_dis.csv", description="")
     SEG_CACHE_PATH: str = Field(default="/seg_cache/", description="")
     RECOMMEND_PATH_PREFIX: str = Field(default="/app/service/recommend/", description="")
+    SERVE_PORT: int = Field(default=2010, description="")
 
 
 settings = Settings()
@@ -117,39 +125,41 @@ KEYPOINT_RESULT_TABLE_FIELD_SET = ('neckline_left', 'neckline_right', 'shoulder_
 MILVUS_TABLE_KEYPOINT = "keypoint_cache_2"
 
 # ollama 地址
-OLLAMA_URL = "http://10.1.1.240:11434/api/embeddings"
+OLLAMA_URL = f"http://{settings.A6000_SERVICE_HOST}:11434/api/embeddings"
 
 """Triton Server Config"""
 # Design
-DESIGN_MODEL_URL = '10.1.1.240:10000'
+DESIGN_MODEL_URL = f'{settings.A6000_SERVICE_HOST}:10000'
 DESIGN_MODEL_NAME = 'seg_knet'
+# Seg Product
+SEG_PRODUCT_MODEL_URL = f'{settings.B_4_X_4090_SERVICE_HOST}:30000'
 # Generate Image
-GI_MODEL_URL = '10.1.1.240:10061'
+GI_MODEL_URL = f'{settings.A6000_SERVICE_HOST}:10061'
 GI_MODEL_NAME = 'flux'
 # Generate Single Logo
-GSL_MODEL_URL = '10.1.1.243:10041'
+GSL_MODEL_URL = f'{settings.B_4_X_4090_SERVICE_HOST}:10041'
 GSL_MODEL_NAME = 'stable_diffusion_xl_transparent'
 # Generate Product （整套和单品）
-GPI_MODEL_URL = '10.1.1.243:10051'
+GPI_MODEL_URL = f'{settings.B_4_X_4090_SERVICE_HOST}:10051'
 GPI_MODEL_NAME_OVERALL = 'diffusion_ensemble_all'
 GPI_MODEL_NAME_SINGLE = 'stable_diffusion_1_5_cnet'
 
 # 以下停用中...*************
 # 多视角生成
-GMV_MODEL_URL = '10.1.1.243:10081'
+GMV_MODEL_URL = f'{settings.B_4_X_4090_SERVICE_HOST}:10081'
 GMV_MODEL_NAME = 'multi_view'
 # 超分
 SR_MODEL_NAME = "super_resolution"
-SR_TRITON_URL = "10.1.1.240:10031"
+SR_TRITON_URL = f"{settings.A6000_SERVICE_HOST}:10031"
 # 打光
-GRI_MODEL_URL = '10.1.1.240:10051'
+GRI_MODEL_URL = f'{settings.A6000_SERVICE_HOST}:10051'
 GRI_MODEL_NAME_OVERALL = 'diffusion_relight_ensemble'
 GRI_MODEL_NAME_SINGLE = 'stable_diffusion_1_5_relight'
 # agent 图片生成
-FAST_GI_MODEL_URL = '10.1.1.243:10011'
+FAST_GI_MODEL_URL = f'{settings.B_4_X_4090_SERVICE_HOST}:10011'
 FAST_GI_MODEL_NAME = 'stable_diffusion_xl'
 # 图转视频 triton版
-PT_MODEL_URL = '10.1.1.243:10061'
+PT_MODEL_URL = f'{settings.B_4_X_4090_SERVICE_HOST}:10061'
 
 # *************
 

app/main.py

@@ -16,7 +16,7 @@ from fastapi.responses import JSONResponse
 
 from app.api.api_route import router
 from app.core.config import settings
-from app.core.record_api_count import count_api_calls
+# from app.core.record_api_count import count_api_calls
 from app.schemas.response_template import ResponseModel
 from logging_env import LOGGER_CONFIG_DICT
 from dotenv import load_dotenv
@@ -48,7 +48,7 @@ def get_application() -> FastAPI:
         allow_methods=["*"],
         allow_headers=["*"],
     )
-    application.middleware("http")(count_api_calls)
+    # application.middleware("http")(count_api_calls)
     application.include_router(router=router)
     return application
 

app/schemas/design.py

@@ -1,4 +1,16 @@
-from pydantic import BaseModel
+from typing import List, Optional
+
+from pydantic import BaseModel, Field
+
+
+class SAMRequestModel(BaseModel):
+    bucket: str = Field(..., description="minio bucket name ")
+    object_name: str = Field(..., description="minio object name ")
+    image_path: str = Field(..., description="图片路径，必填字段")
+    type: str = Field(..., description="推理类型，必填字段")
+    points: Optional[List[List[float]]] | None = None
+    labels: Optional[List[int]] | None = None
+    box: Optional[List[int]] | None = None
 
 
 class DesignModel(BaseModel):

app/schemas/generate_image.py

@@ -1,6 +1,6 @@
-from typing import List
+from typing import List, Optional
 
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 
 
 class GenerateMultiViewModel(BaseModel):
@@ -8,6 +8,17 @@ class GenerateMultiViewModel(BaseModel):
     image_url: str
 
 
+class GenerateImageFlux2KleinModel(BaseModel):
+    bucket_name: str = Field(..., description="OSS桶名，不传则为None")
+    object_name: str = Field(..., description="OSS对象名（文件路径），不传则为None")
+    # input_image_paths: Optional[List[str]] = Field(default=[], description="输入图片路径列表")
+    width: Optional[int] = Field(default=1024, description="图片宽度，默认512像素")
+    height: Optional[int] = Field(default=1024, description="图片高度，默认512像素")
+    prompt: Optional[str] = Field(default="", description="文本提示词，用于模型推理等场景")
+    steps: Optional[int] = Field(default=4, description="推理步数，控制模型生成过程的迭代次数")
+    guidance: Optional[float] = Field(default=4.0, description="引导系数，调节提示词对生成结果的影响程度")
+
+
 class GenerateImageModel(BaseModel):
     tasks_id: str
     prompt: str
@@ -24,6 +35,13 @@ class GenerateSingleLogoImageModel(BaseModel):
     seed: str
 
 
+class GenerateSloganImageModel(BaseModel):
+    num_point: int
+    tasks_id: str
+    prompt: str
+    image_url: str
+
+
 class GenerateProductImageModel(BaseModel):
     tasks_id: str
     prompt: str
@@ -32,6 +50,13 @@ class GenerateProductImageModel(BaseModel):
     product_type: str
 
 
+class Flux2ToProductImgModel(BaseModel):
+    tasks_id: str
+    prompt: str
+    image_path: str
+    infer_step: int | None = None
+
+
 class GenerateRelightImageModel(BaseModel):
     tasks_id: str
     prompt: str

app/service/attribute/service_att_recognition.py

@@ -3,7 +3,6 @@
 from pprint import pprint
 
 import cv2
-import mmcv
 import numpy as np
 import pandas as pd
 import torch
@@ -12,6 +11,7 @@ from minio import Minio
 
 from app.core.config import settings, DESIGN_MODEL_URL
 from app.schemas.attribute_retrieve import AttributeRecognitionModel
+from app.service.utils.image_normalize import my_imnormalize
 from app.service.utils.new_oss_client import oss_get_image
 
 minio_client = Minio(settings.MINIO_URL, access_key=settings.MINIO_ACCESS, secret_key=settings.MINIO_SECRET, secure=settings.MINIO_SECURE)
@@ -109,10 +109,9 @@ class AttributeRecognition:
 
     @staticmethod
     def preprocess(img):
-        img = mmcv.imread(img)
         img_scale = (224, 224)
         img = cv2.resize(img, img_scale)
-        img = mmcv.imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
+        img = my_imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
         preprocessed_img = np.expand_dims(img.transpose(2, 0, 1), axis=0)
         return preprocessed_img
 

app/service/attribute/service_category_recognition.py

@@ -10,7 +10,6 @@
 from minio import Minio
 from skimage import transform
 import cv2
-import mmcv
 import numpy as np
 import pandas as pd
 import tritonclient.http as httpclient
@@ -18,6 +17,7 @@ import torch
 
 from app.core.config import settings, DESIGN_MODEL_URL
 from app.schemas.attribute_retrieve import CategoryRecognitionModel
+from app.service.utils.image_normalize import my_imnormalize
 from app.service.utils.new_oss_client import oss_get_image
 
 minio_client = Minio(settings.MINIO_URL, access_key=settings.MINIO_ACCESS, secret_key=settings.MINIO_SECRET, secure=settings.MINIO_SECURE)
@@ -39,11 +39,10 @@ class CategoryRecognition:
 
     @staticmethod
     def preprocess(img):
-        img = mmcv.imread(img)
         # ori_shape = img.shape[:2]
         img_scale = (224, 224)
         img = cv2.resize(img, img_scale)
-        img = mmcv.imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
+        img = my_imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
         preprocessed_img = np.expand_dims(img.transpose(2, 0, 1), axis=0)
         return preprocessed_img
 

app/service/brand_dna/service.py

@@ -1,7 +1,6 @@
 import logging
 
 import cv2
-import mmcv
 import numpy as np
 import pandas as pd
 import torch
@@ -9,11 +8,12 @@ import torch.nn.functional as F
 import tritonclient.http as httpclient
 from minio import Minio
 
-from app.core.config import DESIGN_MODEL_URL
+from app.core.config import DESIGN_MODEL_URL, SEG_PRODUCT_MODEL_URL
 from app.core.config import settings
 from app.schemas.brand_dna import BrandDnaModel
 from app.service.attribute.config import const
 from app.service.utils.generate_uuid import generate_uuid
+from app.service.utils.image_normalize import my_imnormalize
 from app.service.utils.new_oss_client import oss_upload_image, oss_get_image
 
 minio_client = Minio(settings.MINIO_URL, access_key=settings.MINIO_ACCESS, secret_key=settings.MINIO_SECRET, secure=settings.MINIO_SECURE)
@@ -29,7 +29,7 @@ class BrandDna:
         self.attr_type = pd.read_csv(settings.CATEGORY_PATH)
         # self.attr_type = pd.read_csv(r"E:\workspace\trinity_client_aida\app\service\attribute\config\descriptor\category\category_dis.csv")
         self.att_client = httpclient.InferenceServerClient(url=DESIGN_MODEL_URL)
-        self.seg_client = httpclient.InferenceServerClient(url='10.1.1.243:30000')
+        self.seg_client = httpclient.InferenceServerClient(url=SEG_PRODUCT_MODEL_URL)
         self.const = const
         # self.const = local_debug_const
 
@@ -202,7 +202,7 @@ class BrandDna:
     # 服装分割预处理
     @staticmethod
     def seg_product_preprocess(image):
-        img = mmcv.imread(image)
+        img = image
         ori_shape = img.shape[:2]
         img_scale_w, img_scale_h = ori_shape
         if ori_shape[0] > 1024:
@@ -211,9 +211,9 @@ class BrandDna:
             img_scale_h = 1024
         # 如果图片size任意一边 大于 1024， 则会resize 成1024
         if ori_shape != (img_scale_w, img_scale_h):
-            # mmcv.imresize(img, img_scale_h, img_scale_w) # 老代码 引以为戒！哈哈哈~ h和w写反了
+            # my_imnormalize(img, img_scale_h, img_scale_w) # 老代码 引以为戒！哈哈哈~ h和w写反了
             img = cv2.resize(img, (img_scale_h, img_scale_w))
-        img = mmcv.imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
+        img = my_imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
         preprocessed_img = np.expand_dims(img.transpose(2, 0, 1), axis=0)
         return preprocessed_img, ori_shape
 
@@ -227,11 +227,10 @@ class BrandDna:
     # 类别检测模型预处理
     @staticmethod
     def category_preprocess(img):
-        img = mmcv.imread(img)
         # ori_shape = img.shape[:2]
         img_scale = (224, 224)
         img = cv2.resize(img, img_scale)
-        img = mmcv.imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
+        img = my_imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
         preprocessed_img = np.expand_dims(img.transpose(2, 0, 1), axis=0)
         return preprocessed_img
 

app/service/brand_dna/service_generate_brand_info.py

@@ -1,19 +1,10 @@
-import logging
+import uuid
-
+import httpx
-import cv2
-import numpy as np
-import tritonclient.grpc as grpcclient
 from langchain_classic.output_parsers import ResponseSchema, StructuredOutputParser
 from langchain_community.chat_models import ChatTongyi
 from langchain_core.prompts import PromptTemplate
 from minio import Minio
-from tritonclient.utils import np_to_triton_dtype
-
-from app.core.config import GI_MODEL_URL, GI_MODEL_NAME
 from app.schemas.brand_dna import GenerateBrandModel
-from app.service.utils.generate_uuid import generate_uuid
-from app.service.utils.new_oss_client import oss_upload_image
-
 from app.core.config import settings
 
 
@@ -26,14 +17,9 @@ class GenerateBrandInfo:
         # user info init
         self.user_id = request_data.user_id
         self.category = "brand_logo"
-        # generate logo init
-        self.grpc_client = grpcclient.InferenceServerClient(url=GI_MODEL_URL)
-        self.image = np.random.randint(0, 256, (1024, 1024, 3), dtype=np.uint8)
-        self.batch_size = 1
-        self.mode = 'txt2img'
 
         # llm generate brand info init
-        self.model = ChatTongyi(model="qwen2.5-14b-instruct", api_key="sk-7658298c6b99443c98184a5e634fe6ab")
+        self.model = ChatTongyi(model="qwen2.5-14b-instruct", api_key=settings.QWEN_API_KEY)
 
         self.response_schemas = [
             ResponseSchema(name="brand_name", description="Brand name."),
@@ -63,38 +49,20 @@ class GenerateBrandInfo:
         self.generate_logo_prompt = brand_data['brand_logo_prompt']
 
     def generate_brand_logo(self):
-        prompts = [self.generate_logo_prompt] * self.batch_size
+        request_item = {
-        modes = [self.mode] * self.batch_size
+            "bucket_name": "aida-users",
-        images = [self.image.astype(np.float16)] * self.batch_size
+            "object_name": f'{self.user_id}/{self.category}/{uuid.uuid4().hex}.png',
-
+            "prompt": self.generate_logo_prompt,
-        text_obj = np.array(prompts, dtype="object").reshape((-1, 1))
+            "height": 1024,
-        mode_obj = np.array(modes, dtype="object").reshape((-1, 1))
+            "width": 1024
-        image_obj = np.array(images, dtype=np.float16).reshape((-1, 1024, 1024, 3))
+        }
-
+        with httpx.Client(timeout=120) as client:
-        input_text = grpcclient.InferInput("prompt", text_obj.shape, np_to_triton_dtype(text_obj.dtype))
+            resp = client.post(
-        input_image = grpcclient.InferInput("input_image", image_obj.shape, np_to_triton_dtype(image_obj.dtype))
+                f"http://{settings.FLUX2_GEN_IMG_MODEL_URL}/predict",
-        input_mode = grpcclient.InferInput("mode", mode_obj.shape, np_to_triton_dtype(mode_obj.dtype))
+                json=request_item,
-
+            )
-        input_text.set_data_from_numpy(text_obj)
+        result = resp.json()
-        input_image.set_data_from_numpy(image_obj)
+        self.result_data['brand_logo'] = result.get("output_path", "")
-        input_mode.set_data_from_numpy(mode_obj)
-
-        inputs = [input_text, input_image, input_mode]
-        result = self.grpc_client.infer(model_name=GI_MODEL_NAME, inputs=inputs)
-        image = result.as_numpy("generated_image")
-        image_result = cv2.cvtColor(np.squeeze(image.astype(np.uint8)), cv2.COLOR_RGB2BGR)
-        logo_url = self.upload_logo_image(image_result, generate_uuid())
-        self.result_data['brand_logo'] = logo_url
-
-    def upload_logo_image(self, image, object_name):
-        try:
-            _, img_byte_array = cv2.imencode('.jpg', image)
-            object_name = f'{self.user_id}/{self.category}/{object_name}.jpg'
-            oss_upload_image(oss_client=self.minio_client, bucket="aida-users", object_name=object_name, image_bytes=img_byte_array)
-            image_url = f"aida-users/{object_name}"
-            return image_url
-        except Exception as e:
-            logging.warning(f"upload_png_mask runtime exception : {e}")
 
 
 if __name__ == '__main__':

app/service/clothing_seg/service.py

@@ -23,7 +23,7 @@ class ClothingSeg:
     def __init__(self, request_data):
         self.image_data = request_data.image_data
         self.user_id = request_data.user_id
-        self.triton_client = grpcclient.InferenceServerClient(url="10.1.1.243:10071")
+        self.triton_client = grpcclient.InferenceServerClient(url=f"{settings.B_4_X_4090_SERVICE_HOST}:10071")
 
     @RunTime
     def get_result(self):
@@ -139,7 +139,7 @@ def get_bounding_box(mask):
 
 if __name__ == "__main__":
     test_data = ClothingSegModel(
-        user_id=89,
+        user_id="89",
         image_data=[
             # {
             #     "image_url": "test/clothing_seg/dress.jpg",

app/service/comfyui_I2V/pose2v_server.py

@@ -13,7 +13,7 @@ from PIL import Image
 from minio import Minio, S3Error
 from moviepy.video.io.VideoFileClip import VideoFileClip
 
-from app.core.config import  settings
+from app.core.config import settings, PS_RABBITMQ_QUEUES
 from app.schemas.comfyui_i2v import ComfyuiPose2VModel
 from app.service.generate_image.utils.mq import publish_status
 
@@ -622,9 +622,9 @@ class ComfyUIServerPose2V:
 
             # 推送消息
             if not settings.DEBUG:
-                publish_status(json.dumps(self.pose_transform_data), settings.COMFYUI_SERVER_ADDRESS)
+                publish_status(json.dumps(self.pose_transform_data), PS_RABBITMQ_QUEUES)
             logger.info(
-                f" [x] Sent to： {settings.COMFYUI_SERVER_ADDRESS} data：@@@@  {json.dumps(self.pose_transform_data, indent=4)}")
+                f" [x] Sent to： {PS_RABBITMQ_QUEUES} data：@@@@  {json.dumps(self.pose_transform_data, indent=4)}")
 
             return "\n🎉 所有任务完成！"
 

app/service/design_batch/utils/design_ensemble.py

@@ -10,13 +10,13 @@
 import logging
 
 import cv2
-import mmcv
 import numpy as np
 import torch
 import torch.nn.functional as F
 import tritonclient.http as httpclient
 
 from app.core.config import DESIGN_MODEL_URL, DESIGN_MODEL_NAME
+from app.service.utils.image_normalize import my_imnormalize
 
 """
     keypoint
@@ -25,13 +25,13 @@ from app.core.config import DESIGN_MODEL_URL, DESIGN_MODEL_NAME
 
 
 def keypoint_preprocess(img_path):
-    img = mmcv.imread(img_path)
+    img = img_path
     img_scale = (256, 256)
     h, w = img.shape[:2]
     img = cv2.resize(img, img_scale)
     w_scale = img_scale[0] / w
     h_scale = img_scale[1] / h
-    img = mmcv.imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
+    img = my_imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
     preprocessed_img = np.expand_dims(img.transpose(2, 0, 1), axis=0)
     return preprocessed_img, (w_scale, h_scale)
 
@@ -74,7 +74,7 @@ def keypoint_postprocess(output, scale_factor):
 
 # KNet
 def seg_preprocess(img_path):
-    img = mmcv.imread(img_path)
+    img = img_path
     ori_shape = img.shape[:2]
     img_scale_w, img_scale_h = ori_shape
     if ori_shape[0] > 1024:
@@ -83,9 +83,9 @@ def seg_preprocess(img_path):
         img_scale_h = 1024
     # 如果图片size任意一边 大于 1024， 则会resize 成1024
     if ori_shape != (img_scale_w, img_scale_h):
-        # mmcv.imresize(img, img_scale_h, img_scale_w) # 老代码 引以为戒！哈哈哈~ h和w写反了
+        # my_imnormalize(img, img_scale_h, img_scale_w) # 老代码 引以为戒！哈哈哈~ h和w写反了
         img = cv2.resize(img, (img_scale_h, img_scale_w))
-    img = mmcv.imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
+    img = my_imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
     preprocessed_img = np.expand_dims(img.transpose(2, 0, 1), axis=0)
     return preprocessed_img, ori_shape
 

app/service/design_fast/design_generate.py

@@ -6,10 +6,10 @@ import requests
 from minio import Minio
 
 from app.core.config import settings
-from app.service.design_fast.item import BodyItem, TopItem, BottomItem, OthersItem
+from app.service.design_fast.item import BodyItem, TopItem, BottomItem, OthersItem, TopMergeItem, BottomMergeItem, OthersMergeItem
 from app.service.design_fast.utils.organize import organize_body, organize_clothing, organize_others
 from app.service.design_fast.utils.progress import final_progress, update_progress
-from app.service.design_fast.utils.synthesis_item import synthesis, synthesis_single, update_base_size_priority
+from app.service.design_fast.utils.synthesis_item import synthesis, synthesis_single, update_base_size_priority, merge
 from app.service.utils.decorator import RunTime
 
 id_lock = threading.Lock()
@@ -19,22 +19,46 @@ logger = logging.getLogger()
 minio_client = Minio(settings.MINIO_URL, access_key=settings.MINIO_ACCESS, secret_key=settings.MINIO_SECRET, secure=settings.MINIO_SECURE)
 
 
-def process_item(item, basic):
+def process_item(item, basic, design_type):
-    # 处理project中单个item
+    # 1. 定义映射配置
-    if item['type'] == "Body":
+    # key 为 item_type 的小写，value 为对应的处理类
-        body_server = BodyItem(data=item, basic=basic, minio_client=minio_client)
+    DESIGN_MAP = {
-        item_data = body_server.process()
+        'body': BodyItem,
-    elif item['type'].lower() in ['blouse', 'outwear', 'dress', 'tops']:
+        'blouse': TopItem, 'outwear': TopItem,
-        top_server = TopItem(data=item, basic=basic, minio_client=minio_client)
+        'dress': TopItem, 'tops': TopItem,
-        item_data = top_server.process()
+        'skirt': BottomItem, 'trousers': BottomItem,
-    elif item['type'].lower() in ['skirt', 'trousers', 'bottoms']:
+        'bottoms': BottomItem,
-        bottom_server = BottomItem(data=item, basic=basic, minio_client=minio_client)
+        'others': OthersItem
-        item_data = bottom_server.process()
+    }
-    elif item['type'].lower() in ['others']:
+
-        bottom_server = OthersItem(data=item, basic=basic, minio_client=minio_client)
+    MERGE_MAP = {
-        item_data = bottom_server.process()
+        'body_merge': BodyItem,
+        'blouse_merge': TopMergeItem, 'outwear_merge': TopMergeItem,
+        'dress_merge': TopMergeItem, 'tops_merge': TopMergeItem,
+        'skirt_merge': BottomMergeItem, 'trousers_merge': BottomMergeItem,
+        'bottoms_merge': BottomMergeItem,
+        'others_merge': OthersMergeItem
+    }
+
+    # 2. 根据 design_type 选择映射表
+    mapping = MERGE_MAP if design_type == 'merge' else DESIGN_MAP
+
+    if design_type == 'merge':
+        item_type_key = f"{item['type'].lower()}_merge"
+    elif design_type == 'default':
+        item_type_key = item['type'].lower()
     else:
-        raise NotImplementedError(f"Item type {item['type']} not implemented")
+        item_type_key = item['type'].lower()
+
+    handler_class = mapping.get(item_type_key)
+
+    if not handler_class:
+        raise NotImplementedError(f"Item type {item['type']} not implemented for design_type={design_type}")
+
+    # 4. 统一实例化并执行
+    # 注意：这里假设所有 Item 类构造函数签名一致
+    server = handler_class(data=item, basic=basic, minio_client=minio_client)
+    item_data = server.process()
     return item_data
 
 
@@ -44,7 +68,7 @@ def process_layer(item, layers):
         body_layer = organize_body(item)
         layers.append(body_layer)
         return item['body_image'].size
-    elif item['name'] == 'others':
+    elif item['name'] in ['others', 'others_merge']:
         front_layer, back_layer = organize_others(item)
         layers.append(front_layer)
         layers.append(back_layer)
@@ -70,10 +94,11 @@ def design_generate(request_data):
         nonlocal active_threads
         basic = object['basic']
         items_response = {'layers': [], 'objectSign': object['objectSign'] if 'objectSign' in object.keys() else ""}
+        design_type = basic.get('design_type', "default")
         if basic['single_overall'] == "overall":
             item_results = []
             for item in object['items']:
-                item_results.append(process_item(item, basic))
+                item_results.append(process_item(item, basic, design_type))
             layers = []
             for item in item_results:
                 process_layer(item, layers)
@@ -93,12 +118,19 @@ def design_generate(request_data):
                     'image_url': lay['image_url'] if 'image_url' in lay.keys() else None,
                     'pattern_overall_image_url': lay['pattern_overall_image_url'] if 'pattern_overall_image_url' in lay.keys() else None,
                     'pattern_print_image_url': lay['pattern_print_image_url'] if 'pattern_print_image_url' in lay.keys() else None,
-
+                    'transpose': lay.get('transpose', None),
+                    'rotate': lay.get('rotate', None),
                     # 'back_perspective_url': lay['back_perspective_url'] if 'back_perspective_url' in lay.keys() else None,
                 })
-            items_response['synthesis_url'] = synthesis(layers, new_size, basic)
+            if basic.get('design_type') == 'default':
+                items_response['synthesis_url'] = synthesis(layers, new_size, basic)
+            elif basic.get('design_type') == 'merge':
+                items_response['synthesis_url'] = merge(layers, new_size, basic)
+            else:
+                items_response['synthesis_url'] = synthesis(layers, new_size, basic)
+
         else:
-            item_result = process_item(object['items'][0], basic)
+            item_result = process_item(object['items'][0], basic, design_type)
             items_response['layers'].append({
                 'image_category': f"{item_result['name']}_front",
                 'image_size': item_result['back_image'].size if item_result['back_image'] else None,
@@ -152,6 +184,7 @@ def design_generate_v2(request_data):
 
     def process_object(object, callback_url):
         basic = object['basic']
+        design_type = basic.get('design_type', "default")
         items_response = {
             'layers': [],
             'objectSign': object['objectSign'] if 'objectSign' in object.keys() else "",
@@ -160,7 +193,7 @@ def design_generate_v2(request_data):
         if basic['single_overall'] == "overall":
             item_results = []
             for item in object['items']:
-                item_results.append(process_item(item, basic))
+                item_results.append(process_item(item, basic, design_type))
             layers = []
             for item in item_results:
                 process_layer(item, layers)
@@ -185,7 +218,7 @@ def design_generate_v2(request_data):
                 })
             items_response['synthesis_url'] = synthesis(layers, new_size, basic)
         else:
-            item_result = process_item(object['items'][0], basic)
+            item_result = process_item(object['items'][0], basic, design_type)
             items_response['layers'].append({
                 'image_category': f"{item_result['name']}_front",
                 'image_size': item_result['back_image'].size if item_result['back_image'] else None,

app/service/design_fast/item.py

@@ -7,6 +7,7 @@ class BaseItem:
         self.result['name'] = data['type'].lower()
         self.result.pop("type")
         self.result.update(basic)
+        self.result['design_type'] = basic.get('design_type', None)
 
 
 class OthersItem(BaseItem):
@@ -14,10 +15,7 @@ class OthersItem(BaseItem):
         super().__init__(data, basic)
         self.Others_pipeline = [
             LoadImage(minio_client),
-            # KeyPoint(),
-            # ContourDetection(),
             Segmentation(minio_client),
-            # BackPerspective(minio_client),
             Color(minio_client),
             NoSegPrintPainting(minio_client),
             PrintPainting(minio_client),
@@ -74,6 +72,65 @@ class BottomItem(BaseItem):
         return self.result
 
 
+"""merge"""
+
+
+class OthersMergeItem(BaseItem):
+    def __init__(self, data, basic, minio_client):
+        super().__init__(data, basic)
+        self.Others_pipeline = [
+            LoadImage(minio_client),
+            # KeyPoint(),
+            # ContourDetection(),
+            Segmentation(minio_client),
+            # BackPerspective(minio_client),
+            Color(minio_client),
+            # NoSegPrintPainting(minio_client),
+            # PrintPainting(minio_client),
+            Scaling(),
+            Split(minio_client)
+        ]
+
+    def process(self):
+        for item in self.Others_pipeline:
+            self.result = item(self.result)
+        return self.result
+
+
+class TopMergeItem(BaseItem):
+    def __init__(self, data, basic, minio_client):
+        super().__init__(data, basic)
+        self.top_pipeline = [
+            LoadImage(minio_client),
+            KeyPoint(),
+            Segmentation(minio_client),
+            Scaling(),
+            Split(minio_client)
+        ]
+
+    def process(self):
+        for item in self.top_pipeline:
+            self.result = item(self.result)
+        return self.result
+
+
+class BottomMergeItem(BaseItem):
+    def __init__(self, data, basic, minio_client):
+        super().__init__(data, basic)
+        self.bottom_pipeline = [
+            LoadImage(minio_client),
+            KeyPoint(),
+            Segmentation(minio_client),
+            Scaling(),
+            Split(minio_client)
+        ]
+
+    def process(self):
+        for item in self.bottom_pipeline:
+            self.result = item(self.result)
+        return self.result
+
+
 class BodyItem(BaseItem):
     def __init__(self, data, basic, minio_client):
         super().__init__(data, basic)

app/service/design_fast/pipeline/keypoint.py

@@ -1,7 +1,7 @@
 import logging
 
 import numpy as np
-from pymilvus import MilvusClient
+# from pymilvus import MilvusClient
 
 from app.core.config import KEYPOINT_RESULT_TABLE_FIELD_SET, MILVUS_TABLE_KEYPOINT, settings
 from app.service.design_fast.utils.design_ensemble import get_keypoint_result
@@ -54,63 +54,64 @@ class KeyPoint:
              "keypoint_vector": result.tolist()
              }
         ]
-        try:
+        return dict(zip(KEYPOINT_RESULT_TABLE_FIELD_SET, result.reshape(12, 2).astype(int).tolist()))
-            client = MilvusClient(uri=settings.MILVUS_URL, token=settings.MILVUS_TOKEN, db_name=settings.MILVUS_ALIAS)
-            client.upsert(collection_name=MILVUS_TABLE_KEYPOINT, data=data)
-            client.close()
-            return dict(zip(KEYPOINT_RESULT_TABLE_FIELD_SET, result.reshape(12, 2).astype(int).tolist()))
-        except Exception as e:
-            logger.info(f"save keypoint cache milvus error : {e}")
-            return dict(zip(KEYPOINT_RESULT_TABLE_FIELD_SET, result.reshape(12, 2).astype(int).tolist()))
 
-    @staticmethod
+        # try:
-    def update_keypoint_cache(keypoint_id, infer_result, search_result, site):
+        #     client = MilvusClient(uri=settings.MILVUS_URL, token=settings.MILVUS_TOKEN, db_name=settings.MILVUS_ALIAS)
-        if site == "up":
+        #     client.upsert(collection_name=MILVUS_TABLE_KEYPOINT, data=data)
-            # 需要的是up 即推理出来的是up 那么查询的就是down
+        #     client.close()
-            result = np.concatenate([infer_result.flatten(), search_result[-4:]])
+        # except Exception as e:
-        else:
+        #     logger.info(f"save keypoint cache milvus error : {e}")
-            # 需要的是down 即推理出来的是down 那么查询的就是up
+        #     return dict(zip(KEYPOINT_RESULT_TABLE_FIELD_SET, result.reshape(12, 2).astype(int).tolist()))
-            result = np.concatenate([search_result[:20], infer_result.flatten()])
-        data = [
-            {"keypoint_id": keypoint_id,
-             "keypoint_site": "all",
-             "keypoint_vector": result.tolist()
-             }
-        ]
 
-        try:
+    # @staticmethod
-            client = MilvusClient(uri=settings.MILVUS_URL, token=settings.MILVUS_TOKEN, db_name=settings.MILVUS_ALIAS)
+    # def update_keypoint_cache(keypoint_id, infer_result, search_result, site):
-            client.upsert(
+    #     if site == "up":
-                collection_name=MILVUS_TABLE_KEYPOINT,
+    #         # 需要的是up 即推理出来的是up 那么查询的就是down
-                data=data
+    #         result = np.concatenate([infer_result.flatten(), search_result[-4:]])
-            )
+    #     else:
-            return dict(zip(KEYPOINT_RESULT_TABLE_FIELD_SET, result.reshape(12, 2).astype(int).tolist()))
+    #         # 需要的是down 即推理出来的是down 那么查询的就是up
-        except Exception as e:
+    #         result = np.concatenate([search_result[:20], infer_result.flatten()])
-            logger.info(f"save keypoint cache milvus error : {e}")
+    #     data = [
-            return dict(zip(KEYPOINT_RESULT_TABLE_FIELD_SET, result.reshape(12, 2).astype(int).tolist()))
+    #         {"keypoint_id": keypoint_id,
+    #          "keypoint_site": "all",
+    #          "keypoint_vector": result.tolist()
+    #          }
+    #     ]
+    #
+    #     try:
+    #         client = MilvusClient(uri=settings.MILVUS_URL, token=settings.MILVUS_TOKEN, db_name=settings.MILVUS_ALIAS)
+    #         client.upsert(
+    #             collection_name=MILVUS_TABLE_KEYPOINT,
+    #             data=data
+    #         )
+    #         return dict(zip(KEYPOINT_RESULT_TABLE_FIELD_SET, result.reshape(12, 2).astype(int).tolist()))
+    #     except Exception as e:
+    #         logger.info(f"save keypoint cache milvus error : {e}")
+    #         return dict(zip(KEYPOINT_RESULT_TABLE_FIELD_SET, result.reshape(12, 2).astype(int).tolist()))
 
-    @RunTime
+    # @RunTime
-    def keypoint_cache(self, result, site):
+    # def keypoint_cache(self, result, site):
-        try:
+    #     try:
-            client = MilvusClient(uri=settings.MILVUS_URL, token=settings.MILVUS_TOKEN, db_name=settings.MILVUS_ALIAS)
+    #         client = MilvusClient(uri=settings.MILVUS_URL, token=settings.MILVUS_TOKEN, db_name=settings.MILVUS_ALIAS)
-            keypoint_id = result['image_id']
+    #         keypoint_id = result['image_id']
-            res = client.query(
+    #         res = client.query(
-                collection_name=MILVUS_TABLE_KEYPOINT,
+    #             collection_name=MILVUS_TABLE_KEYPOINT,
-                # ids=[keypoint_id],
+    #             # ids=[keypoint_id],
-                filter=f"keypoint_id == {keypoint_id}",
+    #             filter=f"keypoint_id == {keypoint_id}",
-                output_fields=['keypoint_vector', 'keypoint_site']
+    #             output_fields=['keypoint_vector', 'keypoint_site']
-            )
+    #         )
-            if len(res) == 0:
+    #         if len(res) == 0:
-                # 没有结果 直接推理拿结果 并保存
+    #             # 没有结果 直接推理拿结果 并保存
-                keypoint_infer_result, site = self.infer_keypoint_result(result)
+    #             keypoint_infer_result, site = self.infer_keypoint_result(result)
-                return self.save_keypoint_cache(result['image_id'], keypoint_infer_result, site)
+    #             return self.save_keypoint_cache(result['image_id'], keypoint_infer_result, site)
-            elif res[0]["keypoint_site"] == "all" or res[0]["keypoint_site"] == site:
+    #         elif res[0]["keypoint_site"] == "all" or res[0]["keypoint_site"] == site:
-                # 需要的类型和查询的类型一致，或者查询的类型为all 则直接返回查询的结果
+    #             # 需要的类型和查询的类型一致，或者查询的类型为all 则直接返回查询的结果
-                return dict(zip(KEYPOINT_RESULT_TABLE_FIELD_SET, np.array(res[0]['keypoint_vector']).astype(int).reshape(12, 2).tolist()))
+    #             return dict(zip(KEYPOINT_RESULT_TABLE_FIELD_SET, np.array(res[0]['keypoint_vector']).astype(int).reshape(12, 2).tolist()))
-            elif res[0]["keypoint_site"] != site:
+    #         elif res[0]["keypoint_site"] != site:
-                # 需要的类型和查询到的不一致，则更新类型为all
+    #             # 需要的类型和查询到的不一致，则更新类型为all
-                keypoint_infer_result, site = self.infer_keypoint_result(result)
+    #             keypoint_infer_result, site = self.infer_keypoint_result(result)
-                return self.update_keypoint_cache(result["image_id"], keypoint_infer_result, res[0]['keypoint_vector'], site)
+    #             return self.update_keypoint_cache(result["image_id"], keypoint_infer_result, res[0]['keypoint_vector'], site)
-        except Exception as e:
+    #     except Exception as e:
-            logger.info(f"search keypoint cache milvus error {e}")
+    #         logger.info(f"search keypoint cache milvus error {e}")
-            return False
+    #         return False

app/service/design_fast/pipeline/loading.py

@@ -35,15 +35,9 @@ class LoadImage:
         return cls.name
 
     def __call__(self, result):
+        if result.get("merge_image_path"):
+            result['merge_image'], _ = self.read_image(result['merge_image_path'])
         result['image'], result['pre_mask'] = self.read_image(result['path'])
-        # if 'extract_lines' in result.keys():
-        #     if result['extract_lines']:
-        #         result['gray'] = self.get_lines(cv2.cvtColor(result['image'], cv2.COLOR_BGR2GRAY), result['path'])
-        #     else:
-        #         result['gray'] = cv2.cvtColor(result['image'], cv2.COLOR_BGR2GRAY)
-        # else:
-        #     result['gray'] = cv2.cvtColor(result['image'], cv2.COLOR_BGR2GRAY)
-
         result['gray'] = self.get_lines(cv2.cvtColor(result['image'], cv2.COLOR_BGR2GRAY))
         result['keypoint'] = self.get_keypoint(result['name'])
         result['img_shape'] = result['image'].shape
@@ -61,21 +55,6 @@ class LoadImage:
         mask = skeleton
         result = np.ones_like(img) * 255
         result[mask] = img[mask]
-
-        # 步骤2：细化边缘（可选，让线条更干净）
-        # kernel = np.ones((1, 1), np.uint8)
-        # clean = cv2.morphologyEx(binary, cv2.MORPH_OPEN, kernel)
-
-        # thinned = cv2.ximgproc.thinning(binary, thinningType=cv2.ximgproc.THINNING_ZHANGSUEN)  # thinning算法细化线条
-        # mask = thinned > 0
-        # result = np.ones_like(img) * 255
-        # result[mask] = img[mask]
-
-        # 步骤3：反转回 白底黑线
-        # lines = cv2.bitwise_not(thinned)
-        # cv2.imwrite(os.path.join('/home/user/PycharmProjects/trinity_client_aida/test/lines_original_result_5', f"Original_{path.replace('/', '-')}.png"), img)
-        # cv2.imwrite(os.path.join('/home/user/PycharmProjects/trinity_client_aida/test/lines_original_result_5', f"Line_{path.replace('/', '-')}.png"), result)
-
         return result
 
     def read_image(self, image_path):
@@ -96,19 +75,19 @@ class LoadImage:
 
     @staticmethod
     def get_keypoint(name):
-        if name == 'blouse' or name == 'outwear' or name == 'dress' or name == 'tops':
+        if name in ['blouse', 'outwear', 'dress', 'tops', 'blouse_merge', 'outwear_merge', 'dress_merge', 'tops_merge']:
             keypoint = 'shoulder'
-        elif name == 'trousers' or name == 'skirt' or name == 'bottoms':
+        elif name in ['trousers', 'skirt', 'bottoms', 'trousers_merge', 'skirt_merge', 'bottoms_merge']:
             keypoint = 'waistband'
-        elif name == 'bag':
+        elif name in ['bag', 'bag_merge']:
             keypoint = 'hand_point'
-        elif name == 'shoes':
+        elif name in ['shoes', 'shoes_merge']:
             keypoint = 'toe'
-        elif name == 'hairstyle':
+        elif name in ['hairstyle', 'hairstyle_merge']:
             keypoint = 'head_point'
-        elif name == 'earring':
+        elif name in ['earring', 'earring_merge']:
             keypoint = 'ear_point'
-        elif name == 'others':
+        elif name in ['others', 'others_merge']:
             keypoint = "others"
         else:
             raise KeyError(f"{name} does not belong to item category list: blouse, outwear, dress, trousers, skirt, "

app/service/design_fast/pipeline/no_seg_print_painting.py

@@ -9,32 +9,27 @@ from app.service.utils.new_oss_client import oss_get_image
 
 class NoSegPrintPainting:
     def __init__(self, minio_client):
-        self.random_seed = random.randint(0, 1000)
         self.minio_client = minio_client
 
     def __call__(self, result):
-        single_print = result['print']['single']
+        # single_print = [result['print']['single']]
         overall_print = result['print']['overall']
-        element_print = result['print']['element']
+        # element_print = result['print']['element'
+
+        single_print = None
+        element_print = None
+
         result['single_image'] = None
         result['print_image'] = None
 
         if overall_print['print_path_list']:
             painting_dict = {'dim_image_h': result['pattern_image'].shape[0], 'dim_image_w': result['pattern_image'].shape[1]}
-            if "print_angle_list" in overall_print.keys() and overall_print['print_angle_list'][0] != 0:
+            # 获取平铺 + 旋转 的overall print
-                painting_dict = self.painting_collection(painting_dict, overall_print, print_trigger=True)
+            painting_dict = self.painting_collection(painting_dict, overall_print)
-                painting_dict['tile_print'] = self.rotate_crop_image(img=painting_dict['tile_print'], angle=-overall_print['print_angle_list'][0], crop=True)
-                painting_dict['mask_inv_print'] = self.rotate_crop_image(img=painting_dict['mask_inv_print'], angle=-overall_print['print_angle_list'][0], crop=True)
-
-                # resize 到sketch大小
-                painting_dict['tile_print'] = self.resize_and_crop(img=painting_dict['tile_print'], target_width=painting_dict['dim_image_w'], target_height=painting_dict['dim_image_h'])
-                painting_dict['mask_inv_print'] = self.resize_and_crop(img=painting_dict['mask_inv_print'], target_width=painting_dict['dim_image_w'], target_height=painting_dict['dim_image_h'])
-            else:
-                painting_dict = self.painting_collection(painting_dict, overall_print, print_trigger=True, is_single=False)
             result['no_seg_sketch_overall'] = result['no_seg_sketch_print'] = self.printpaint(result, painting_dict, print_=True)
-            result['pattern_image'] = result['no_seg_sketch_overall']
+            # result['pattern_image'] = result['no_seg_sketch_overall']
 
-        if single_print['print_path_list']:
+        if single_print:
             print_background = np.zeros((result['pattern_image'].shape[0], result['pattern_image'].shape[1], 3), dtype=np.uint8)
             mask_background = np.zeros((result['pattern_image'].shape[0], result['pattern_image'].shape[1], 3), dtype=np.uint8)
             for i in range(len(single_print['print_path_list'])):
@@ -74,7 +69,7 @@ class NoSegPrintPainting:
                 single_image = cv2.add(tmp1, tmp2)
                 result['no_seg_sketch_print'] = single_image
 
-        if element_print['element_path_list']:
+        if element_print:
             print_background = np.zeros((result['final_image'].shape[0], result['final_image'].shape[1], 3), dtype=np.uint8)
             mask_background = np.zeros((result['final_image'].shape[0], result['final_image'].shape[1], 3), dtype=np.uint8)
             for i in range(len(element_print['element_path_list'])):
@@ -151,7 +146,6 @@ class NoSegPrintPainting:
             temp_fg = np.expand_dims(result['mask'], axis=2).repeat(3, axis=2)
             tmp2 = (result['final_image'] * (temp_fg / 255)).astype(np.uint8)
             result['no_seg_sketch_print'] = cv2.add(tmp1, tmp2)
-
         return result
 
     @staticmethod
@@ -166,26 +160,23 @@ class NoSegPrintPainting:
         print_background = img1_bg + img2_fg
         return print_background
 
-    def painting_collection(self, painting_dict, print_dict, print_trigger=False, is_single=False):
+    def painting_collection(self, painting_dict, print_dict):
-        if print_trigger:
+        print_ = self.get_print(print_dict)
-            print_ = self.get_print(print_dict)
+        painting_dict['location'] = print_['location']
-            painting_dict['Trigger'] = not is_single
+        dim_max = max(painting_dict['dim_image_h'], painting_dict['dim_image_w'])
-            painting_dict['location'] = print_['location']
+        dim_pattern = (int(dim_max * print_['scale'] / 5), int(dim_max * print_['scale'] / 5))
-            single_mask_inv_print = self.get_mask_inv(print_['image'])
+        gap = print_dict.get('gap', [[0, 0]])[0]
-            dim_max = max(painting_dict['dim_image_h'], painting_dict['dim_image_w'])
+        painting_dict['tile_print'], painting_dict['mask_inv_print'] = tile_image(pattern=print_['image'],
-            dim_pattern = (int(dim_max * print_['scale'] / 5), int(dim_max * print_['scale'] / 5))
+                                                                                  mask=print_['mask'],
-            if not is_single:
+                                                                                  dim=dim_pattern,
-                # 如果print 模式为overall 且 有角度的话 ， 组合的print为正方形，方便裁剪
+                                                                                  gap_x=gap[0],
-                if "print_angle_list" in print_dict.keys() and print_dict['print_angle_list'][0] != 0:
+                                                                                  gap_y=gap[1],
-                    painting_dict['mask_inv_print'] = self.tile_image(single_mask_inv_print, dim_pattern, print_['scale'], dim_max, dim_max, painting_dict['location'], trigger=True)
+                                                                                  canvas_h=painting_dict['dim_image_h'],
-                    painting_dict['tile_print'] = self.tile_image(print_['image'], dim_pattern, print_['scale'], dim_max, dim_max, painting_dict['location'], trigger=True)
+                                                                                  canvas_w=painting_dict['dim_image_w'],
-                else:
+                                                                                  location=painting_dict['location'],
-                    painting_dict['mask_inv_print'] = self.tile_image(single_mask_inv_print, dim_pattern, print_['scale'], painting_dict['dim_image_h'], painting_dict['dim_image_w'], painting_dict['location'], trigger=True)
+                                                                                  angle=int(print_.get('print_angle_list', [0])[0]))
-                    painting_dict['tile_print'] = self.tile_image(print_['image'], dim_pattern, print_['scale'], painting_dict['dim_image_h'], painting_dict['dim_image_w'], painting_dict['location'], trigger=True)
+        # painting_dict['mask_inv_print'] = np.zeros(painting_dict['tile_print'].shape[:2], dtype=np.uint8)
-            else:
+        # painting_dict['mask_inv_print'] = self.get_mask_inv(painting_dict['tile_print'])
-                painting_dict['mask_inv_print'] = self.tile_image(single_mask_inv_print, dim_pattern, print_['scale'], painting_dict['dim_image_h'], painting_dict['dim_image_w'], painting_dict['location'])
-                painting_dict['tile_print'] = self.tile_image(print_['image'], dim_pattern, print_['scale'], painting_dict['dim_image_h'], painting_dict['dim_image_w'], painting_dict['location'])
-                painting_dict['dim_print_h'], painting_dict['dim_print_w'] = dim_pattern
         return painting_dict
 
     def tile_image(self, pattern, dim, scale, dim_image_h, dim_image_w, location, trigger=False):
@@ -219,33 +210,32 @@ class NoSegPrintPainting:
 
     @staticmethod
     def printpaint(result, painting_dict, print_=False):
-
+        if print_:
-        if print_ and painting_dict['Trigger']:
             print_mask = cv2.bitwise_and(result['mask'], cv2.bitwise_not(painting_dict['mask_inv_print']))
             img_fg = cv2.bitwise_and(painting_dict['tile_print'], painting_dict['tile_print'], mask=print_mask)
         else:
             print_mask = result['mask']
             img_fg = result['final_image']
-        if print_ and not painting_dict['Trigger']:
+        # if print_ and not painting_dict['Trigger']:
-            index_ = None
+        #     index_ = None
-            try:
+        #     try:
-                index_ = len(painting_dict['location'])
+        #         index_ = len(painting_dict['location'])
-            except:
+        #     except:
-                assert f'there must be parameter of location if choose IfSingle'
+        #         assert f'there must be parameter of location if choose IfSingle'
-
+        #
-            for i in range(index_):
+        #     for i in range(index_):
-                start_h, start_w = int(painting_dict['location'][i][1]), int(painting_dict['location'][i][0])
+        #         start_h, start_w = int(painting_dict['location'][i][1]), int(painting_dict['location'][i][0])
-
+        #
-                length_h = min(start_h + painting_dict['dim_print_h'], img_fg.shape[0])
+        #         length_h = min(start_h + painting_dict['dim_print_h'], img_fg.shape[0])
-                length_w = min(start_w + painting_dict['dim_print_w'], img_fg.shape[1])
+        #         length_w = min(start_w + painting_dict['dim_print_w'], img_fg.shape[1])
-
+        #
-                change_region = img_fg[start_h: length_h, start_w: length_w, :]
+        #         change_region = img_fg[start_h: length_h, start_w: length_w, :]
-                # problem in change_mask
+        #         # problem in change_mask
-                change_mask = print_mask[start_h: length_h, start_w: length_w]
+        #         change_mask = print_mask[start_h: length_h, start_w: length_w]
-                # get real part into change mask
+        #         # get real part into change mask
-                _, change_mask = cv2.threshold(change_mask, 220, 255, cv2.THRESH_BINARY)
+        #         _, change_mask = cv2.threshold(change_mask, 220, 255, cv2.THRESH_BINARY)
-                cv2.bitwise_not(painting_dict['mask_inv_print'])
+        #         cv2.bitwise_not(painting_dict['mask_inv_print'])
-                img_fg[start_h:start_h + painting_dict['dim_print_h'], start_w:start_w + painting_dict['dim_print_w'], :] = change_region
+        #         img_fg[start_h:start_h + painting_dict['dim_print_h'], start_w:start_w + painting_dict['dim_print_w'], :] = change_region
 
         clothes_mask_print = cv2.bitwise_not(print_mask)
 
@@ -267,18 +257,21 @@ class NoSegPrintPainting:
         image = oss_get_image(oss_client=self.minio_client, bucket=bucket_name, object_name=object_name, data_type="PIL")
         # 判断图片格式，如果是RGBA 则贴在一张纯白图片上 防止透明转黑
         if image.mode == "RGBA":
+            mask_pil = image.split()[3]
             new_background = Image.new('RGB', image.size, (255, 255, 255))
             new_background.paste(image, mask=image.split()[3])
             image = new_background
+        else:
+            mask_pil = Image.new('L', image.size, 255)  # L=灰度图，255=纯白
         print_dict['image'] = cv2.cvtColor(np.asarray(image), cv2.COLOR_RGB2BGR)
+        print_dict['mask'] = cv2.threshold(np.array(mask_pil), 127, 255, cv2.THRESH_BINARY)[1]
+
         return print_dict
 
     def crop_image(self, image, image_size_h, image_size_w, location, print_shape):
         print_w = print_shape[1]
         print_h = print_shape[0]
 
-        random.seed(self.random_seed)
-
         # 1.拿到偏移量后和resize后的print宽高取余 得到真正偏移量
         # 偏移量增加2分之print.w 使坐标位于图中间  如果要位于左上角删除+ print_w // 2 即可
         x_offset = print_w - int(location[0][1] % print_w) + print_w // 2
@@ -420,3 +413,118 @@ class NoSegPrintPainting:
             cropped_img = resized_img[start_y:start_y + target_height, :]
 
         return cropped_img
+
+
+def tile_image(pattern, mask, dim, gap_x, gap_y, canvas_h, canvas_w, location, angle=0):
+    """
+    按照指定的 X/Y 间距平铺印花，并支持旋转
+    【修改版】以被平铺图案的【中心】作为平铺基准点
+
+    :param location: [[center_y, center_x]]  → 第一个图案中心的坐标
+    :param angle: 旋转角度 (度数, 逆时针)
+    """
+    # 1. 确保输入是 RGBA
+    if pattern.shape[2] == 3:
+        pattern = cv2.cvtColor(pattern, cv2.COLOR_BGR2BGRA)
+
+    # 2. 缩放与旋转印花
+    resized_p = cv2.resize(pattern, dim, interpolation=cv2.INTER_AREA)
+    rotated_p = rotate_image(resized_p, angle)
+    p_h, p_w = rotated_p.shape[:2]
+
+    # 3. 创建透明单元格（图案放在单元格中心）
+    cell_h = p_h + gap_y
+    cell_w = p_w + gap_x
+
+    unit_cell = np.zeros((cell_h, cell_w, 4), dtype=np.uint8)
+
+    # 计算图案在单元格中的左上角位置（让图案居中）
+    start_y = (cell_h - p_h) // 2
+    start_x = (cell_w - p_w) // 2
+    unit_cell[start_y:start_y + p_h, start_x:start_x + p_w, :] = rotated_p
+
+    # 4. 执行平铺
+    tiles_y = (canvas_h // cell_h) + 3  # 多加一点余量更安全
+    tiles_x = (canvas_w // cell_w) + 3
+    full_tiled = np.tile(unit_cell, (tiles_y, tiles_x, 1))
+
+    # 5. 计算偏移（关键修改：以中心为基准）
+    center_y, center_x = location[0][0], location[0][1]  # 第一个图案的中心位置
+
+    # 计算从哪个位置开始裁剪，才能让中心落在指定坐标
+    offset_y = int((center_y - (p_h // 2)) % cell_h)
+    offset_x = int((center_x - (p_w // 2)) % cell_w)
+
+    tiled_layer = full_tiled[offset_y: offset_y + canvas_h,
+    offset_x: offset_x + canvas_w]
+
+    # 6. 创建纯白色背景并合成（保持你原来的风格）
+    white_background = np.full((canvas_h, canvas_w, 3), 255, dtype=np.uint8)
+
+    tiled_bgr = tiled_layer[:, :, :3]
+    alpha_mask = tiled_layer[:, :, 3] / 255.0
+    alpha_mask = cv2.merge([alpha_mask, alpha_mask, alpha_mask])
+
+    tiled_print = (tiled_bgr * alpha_mask + white_background * (1 - alpha_mask)).astype(np.uint8)
+
+    # ====================== 处理 Mask ======================
+    # Mask 也同样居中处理
+    resized_mask = cv2.resize(mask, dim, interpolation=cv2.INTER_NEAREST)
+    rotated_mask = rotate_image(resized_mask, angle)  # 注意：mask也需要旋转
+
+    unit_mask = np.zeros((cell_h, cell_w), dtype=np.uint8)
+    unit_mask[start_y:start_y + p_h, start_x:start_x + p_w] = rotated_mask
+
+    full_mask_tiled = np.tile(unit_mask, (tiles_y, tiles_x))
+    tiled_mask = full_mask_tiled[offset_y: offset_y + canvas_h,
+    offset_x: offset_x + canvas_w]
+
+    return tiled_print, cv2.bitwise_not(tiled_mask)
+
+
+def rotate_image(image, angle):
+    """
+    旋转图片并保持完整内容（自动扩大画布）
+    """
+    if angle == 0:
+        return image
+
+    (h, w) = image.shape[:2]
+    (cX, cY) = (w // 2, h // 2)
+
+    # 获取旋转矩阵
+    M = cv2.getRotationMatrix2D((cX, cY), angle, 1.0)
+
+    # 计算旋转后新边界的 sine 和 cosine
+    cos = np.abs(M[0, 0])
+    sin = np.abs(M[0, 1])
+
+    # 计算新的画布尺寸
+    nW = int((h * sin) + (w * cos))
+    nH = int((h * cos) + (w * sin))
+
+    # 调整旋转矩阵以考虑平移
+    M[0, 2] += (nW / 2) - cX
+    M[1, 2] += (nH / 2) - cY
+
+    # 执行旋转
+    return cv2.warpAffine(image, M, (nW, nH))
+
+
+def crop_image(image, image_size_h, image_size_w, location, print_shape):
+    print_w = print_shape[1]
+    print_h = print_shape[0]
+
+    # 1.拿到偏移量后和resize后的print宽高取余 得到真正偏移量
+    # 偏移量增加2分之print.w 使坐标位于图中间  如果要位于左上角删除+ print_w // 2 即可
+    x_offset = print_w - int(location[0][1] % print_w) + print_w // 2
+    y_offset = print_h - int(location[0][0] % print_h) + print_h // 2
+
+    # y_offset = int(location[0][0])
+    # x_offset = int(location[0][1])
+
+    if len(image.shape) == 2:
+        image = image[x_offset: x_offset + image_size_h, y_offset: y_offset + image_size_w]
+    elif len(image.shape) == 3:
+        image = image[x_offset: x_offset + image_size_h, y_offset: y_offset + image_size_w, :]
+    return image

app/service/design_fast/pipeline/print_painting.py

@@ -9,14 +9,17 @@ from app.service.utils.new_oss_client import oss_get_image
 
 class PrintPainting:
     def __init__(self, minio_client):
-        self.random_seed = None
         self.minio_client = minio_client
 
     def __call__(self, result):
-        single_print = result['print']['single']
+        # single_print = result['print']['single']
         overall_print = result['print']['overall']
-        element_print = result['print']['element']
+        # element_print = result['print']['element']
-        partial_path = result['print']['partial'] if 'partial' in result['print'] else None
+        # partial_path = result['print']['partial'] if 'partial' in result['print'] else None
+
+        single_print = None
+        element_print = None
+        partial_path = None
         result['single_image'] = None
         result['print_image'] = None
         # TODO 给result['pattern_image'] resize 到resize_scale的大小
@@ -38,24 +41,15 @@ class PrintPainting:
         if overall_print['print_path_list']:
             overall_print['location'][0] = [x * y for x, y in zip(overall_print['location'][0], result['resize_scale'])]
             painting_dict = {'dim_image_h': result['pattern_image'].shape[0], 'dim_image_w': result['pattern_image'].shape[1]}
-            result['print_image'] = result['pattern_image']
+            result['print_image'] = result['pattern_image'].copy()
-            if "print_angle_list" in overall_print.keys() and overall_print['print_angle_list'][0] != 0:
+            # 获取平铺 + 旋转 的overall print
-                painting_dict = self.painting_collection(painting_dict, overall_print, print_trigger=True)
+            painting_dict = self.painting_collection(painting_dict, overall_print)
-                painting_dict['tile_print'] = self.rotate_crop_image(img=painting_dict['tile_print'], angle=-overall_print['print_angle_list'][0], crop=True)
-                painting_dict['mask_inv_print'] = self.rotate_crop_image(img=painting_dict['mask_inv_print'], angle=-overall_print['print_angle_list'][0], crop=True)
-
-                # resize 到sketch大小
-                painting_dict['tile_print'] = self.resize_and_crop(img=painting_dict['tile_print'], target_width=painting_dict['dim_image_w'], target_height=painting_dict['dim_image_h'])
-                painting_dict['mask_inv_print'] = self.resize_and_crop(img=painting_dict['mask_inv_print'], target_width=painting_dict['dim_image_w'], target_height=painting_dict['dim_image_h'])
-            else:
-                painting_dict = self.painting_collection(painting_dict, overall_print, print_trigger=True, is_single=False)
             result['print_image'] = self.printpaint(result, painting_dict, print_=True)
             result['single_image'] = result['final_image'] = result['pattern_image'] = result['print_image']
 
-        if single_print['print_path_list']:
+        if single_print:
             # 2025-9-19 印花调整 印花坐标按照sketch的缩放比调整
             sketch_resize_scale = result['resize_scale']
-
             print_background = np.zeros((result['pattern_image'].shape[0], result['pattern_image'].shape[1], 3), dtype=np.uint8)
             mask_background = np.zeros((result['pattern_image'].shape[0], result['pattern_image'].shape[1], 3), dtype=np.uint8)
             for i in range(len(single_print['print_path_list'])):
@@ -78,75 +72,6 @@ class PrintPainting:
                 print_background = cv2.cvtColor(np.array(source_image_pil), cv2.COLOR_RGBA2BGR)
                 mask_background = cv2.cvtColor(np.array(source_image_pil_mask), cv2.COLOR_RGBA2BGR)
                 ret, mask_background = cv2.threshold(mask_background, 124, 255, cv2.THRESH_BINARY)
-                # else:
-                #     mask = self.get_mask_inv(image)
-                #     mask = np.expand_dims(mask, axis=2)
-                #     mask = cv2.cvtColor(mask, cv2.COLOR_GRAY2BGR)
-                #     mask = cv2.bitwise_not(mask)
-                #
-                #     mask = cv2.resize(mask, (int(result['final_image'].shape[1] * single_print['print_scale_list'][i][0]), int(result['final_image'].shape[0] * single_print['print_scale_list'][i][1])))
-                #     image = cv2.resize(image, (int(result['final_image'].shape[1] * single_print['print_scale_list'][i][0]), int(result['final_image'].shape[0] * single_print['print_scale_list'][i][1])))
-                #     # 旋转后的坐标需要重新算
-                #     rotate_mask, _ = self.img_rotate(mask, single_print['print_angle_list'][i])
-                #     rotate_image, rotated_new_size = self.img_rotate(image, single_print['print_angle_list'][i])
-                #     # x, y = int(result['print']['location'][i][0] - rotated_new_size[0] - (rotate_mask.shape[0] - image.shape[0]) / 2), int(result['print']['location'][i][1] - rotated_new_size[1] - (rotate_mask.shape[1] - image.shape[1]) / 2)
-                #     x, y = int(single_print['location'][i][0] - rotated_new_size[0]), int(single_print['location'][i][1] - rotated_new_size[1])
-                #
-                #     image_x = print_background.shape[1]  # 底图宽
-                #     image_y = print_background.shape[0]  # 底图高
-                #     print_x = rotate_image.shape[1]     #印花宽
-                #     print_y = rotate_image.shape[0]     #印花高
-                #
-                #     #  有bug
-                #     # if x + print_x > image_x:
-                #     #     rotate_image = rotate_image[:, :x + print_x - image_x]
-                #     #     rotate_mask = rotate_mask[:, :x + print_x - image_x]
-                #     # #
-                #     # if y + print_y > image_y:
-                #     #     rotate_image = rotate_image[:y + print_y - image_y]
-                #     #     rotate_mask = rotate_mask[:y + print_y - image_y]
-                #
-                #     #  不能是并行
-                #     # 当前第一轮的if （108以及115）是判断有没有过下界和右界。第二轮的是判断左上有没有超出。 如果这个样子的话，先裁了右边，再左移，region就会有问题
-                #     # 先挪 再判断 最后裁剪
-                #
-                #     # 如果print旋转了 或者 print贴边了 则需要判断 判断左界和上界是否小于0
-                #     if x <= 0: # 如果X轴偏移量小于0，说明印花需要被裁剪至合适大小 或当X轴偏移量大于印花宽度时，裁剪后的印花宽度为0
-                #         rotate_image = rotate_image[:, abs(x):]
-                #         rotate_mask = rotate_mask[:, abs(x):]
-                #         start_x = x = 0
-                #     else:
-                #         start_x = x
-                #
-                #     if y <= 0:  # 如果X轴偏移量大于0，说明印花需要被裁剪至合适大小 或当Y轴偏移量大于印花宽度时，裁剪后的印花宽度为0
-                #         rotate_image = rotate_image[abs(y):, :]
-                #         rotate_mask = rotate_mask[abs(y):, :]
-                #         start_y = y = 0
-                #     else:
-                #         start_y = y
-                #
-                #     # ------------------
-                #     # 如果print-size大于image-size 则需要裁剪print
-                #
-                #     if x + print_x > image_x:
-                #         rotate_image = rotate_image[:, :image_x - x]
-                #         rotate_mask = rotate_mask[:, :image_x - x]
-                #
-                #     if y + print_y > image_y:
-                #         rotate_image = rotate_image[:image_y - y, :]
-                #         rotate_mask = rotate_mask[:image_y - y, :]
-                #
-                #     # mask_background[start_y:y + rotate_mask.shape[0], start_x:x + rotate_mask.shape[1]] = cv2.bitwise_xor(mask_background[start_y:y + rotate_mask.shape[0], start_x:x + rotate_mask.shape[1]], rotate_mask)
-                #     # print_background[start_y:y + rotate_image.shape[0], start_x:x + rotate_image.shape[1]] = cv2.add(print_background[start_y:y + rotate_image.shape[0], start_x:x + rotate_image.shape[1]], rotate_image)
-                #
-                #     # mask_background[start_y:y + rotate_mask.shape[0], start_x:x + rotate_mask.shape[1]] = rotate_mask
-                #     # print_background[start_y:y + rotate_image.shape[0], start_x:x + rotate_image.shape[1]] = rotate_image
-                #     mask_background = self.stack_prin(mask_background, result['pattern_image'], rotate_mask, start_y, y, start_x, x)
-                #     print_background = self.stack_prin(print_background, result['pattern_image'], rotate_image, start_y, y, start_x, x)
-
-                # gray_image = cv2.cvtColor(mask_background, cv2.COLOR_BGR2GRAY)
-                # print_background = cv2.bitwise_and(print_background, print_background, mask=gray_image)
-
                 print_mask = cv2.bitwise_and(result['mask'], cv2.cvtColor(mask_background, cv2.COLOR_BGR2GRAY))
                 img_fg = cv2.bitwise_or(print_background, print_background, mask=print_mask)
                 img_bg = cv2.bitwise_and(result['pattern_image'], result['pattern_image'], mask=cv2.bitwise_not(print_mask))
@@ -163,10 +88,9 @@ class PrintPainting:
                 tmp2 = (result['final_image'] * (temp_fg / 255)).astype(np.uint8)
                 result['single_image'] = cv2.add(tmp1, tmp2)
 
-        if element_print['element_path_list']:
+        if element_print:
             # 2025-9-19 印花调整 印花坐标按照sketch的缩放比调整
             sketch_resize_scale = result['resize_scale']
-
             print_background = np.zeros((result['final_image'].shape[0], result['final_image'].shape[1], 3), dtype=np.uint8)
             mask_background = np.zeros((result['final_image'].shape[0], result['final_image'].shape[1], 3), dtype=np.uint8)
             for i in range(len(element_print['element_path_list'])):
@@ -207,20 +131,6 @@ class PrintPainting:
                     print_x = rotate_image.shape[1]
                     print_y = rotate_image.shape[0]
 
-                    #  有bug
-                    # if x + print_x > image_x:
-                    #     rotate_image = rotate_image[:, :x + print_x - image_x]
-                    #     rotate_mask = rotate_mask[:, :x + print_x - image_x]
-                    # #
-                    # if y + print_y > image_y:
-                    #     rotate_image = rotate_image[:y + print_y - image_y]
-                    #     rotate_mask = rotate_mask[:y + print_y - image_y]
-
-                    #  不能是并行
-                    # 当前第一轮的if （108以及115）是判断有没有过下界和右界。第二轮的是判断左上有没有超出。 如果这个样子的话，先裁了右边，再左移，region就会有问题
-                    # 先挪 再判断 最后裁剪
-
-                    # 如果print旋转了 或者 print贴边了 则需要判断 判断左界和上界是否小于0
                     if x <= 0:
                         rotate_image = rotate_image[:, -x:]
                         rotate_mask = rotate_mask[:, -x:]
@@ -235,9 +145,6 @@ class PrintPainting:
                     else:
                         start_y = y
 
-                    # ------------------
-                    # 如果print-size大于image-size 则需要裁剪print
-
                     if x + print_x > image_x:
                         rotate_image = rotate_image[:, :image_x - x]
                         rotate_mask = rotate_mask[:, :image_x - x]
@@ -246,11 +153,6 @@ class PrintPainting:
                         rotate_image = rotate_image[:image_y - y, :]
                         rotate_mask = rotate_mask[:image_y - y, :]
 
-                    # mask_background[start_y:y + rotate_mask.shape[0], start_x:x + rotate_mask.shape[1]] = cv2.bitwise_xor(mask_background[start_y:y + rotate_mask.shape[0], start_x:x + rotate_mask.shape[1]], rotate_mask)
-                    # print_background[start_y:y + rotate_image.shape[0], start_x:x + rotate_image.shape[1]] = cv2.add(print_background[start_y:y + rotate_image.shape[0], start_x:x + rotate_image.shape[1]], rotate_image)
-
-                    # mask_background[start_y:y + rotate_mask.shape[0], start_x:x + rotate_mask.shape[1]] = rotate_mask
-                    # print_background[start_y:y + rotate_image.shape[0], start_x:x + rotate_image.shape[1]] = rotate_image
                     mask_background = self.stack_prin(mask_background, result['pattern_image'], rotate_mask, start_y, y, start_x, x)
                     print_background = self.stack_prin(print_background, result['pattern_image'], rotate_image, start_y, y, start_x, x)
 
@@ -298,12 +200,8 @@ class PrintPainting:
                 ret, mask_background = cv2.threshold(mask_background, 124, 255, cv2.THRESH_BINARY)
             print_mask = cv2.bitwise_and(result['mask'], cv2.cvtColor(mask_background, cv2.COLOR_BGR2GRAY))
             img_fg = cv2.bitwise_or(print_background, print_background, mask=print_mask)
-            # TODO element  丢失信息
             three_channel_image = cv2.merge([cv2.bitwise_not(print_mask), cv2.bitwise_not(print_mask), cv2.bitwise_not(print_mask)])
             img_bg = cv2.bitwise_and(result['final_image'], three_channel_image)
-            # mask_mo = np.expand_dims(print_mask, axis=2).repeat(3, axis=2)
-            # gray_mo = np.expand_dims(result['gray'], axis=2).repeat(3, axis=2)
-            # img_fg = (img_fg * (mask_mo / 255) * (gray_mo / 255)).astype(np.uint8)
             result['final_image'] = cv2.add(img_bg, img_fg)
             canvas = np.full_like(result['final_image'], 255)
             temp_bg = np.expand_dims(cv2.bitwise_not(result['mask']), axis=2).repeat(3, axis=2)
@@ -325,27 +223,21 @@ class PrintPainting:
         print_background = img1_bg + img2_fg
         return print_background
 
-    def painting_collection(self, painting_dict, print_dict, print_trigger=False, is_single=False):
+    def painting_collection(self, painting_dict, print_dict):
-        if print_trigger:
+        print_ = self.get_print(print_dict)
-            print_ = self.get_print(print_dict)
+        painting_dict['location'] = print_['location']
-            painting_dict['Trigger'] = not is_single
+        dim_max = max(painting_dict['dim_image_h'], painting_dict['dim_image_w'])
-            painting_dict['location'] = print_['location']
+        dim_pattern = (int(dim_max * print_['scale'] / 5), int(dim_max * print_['scale'] / 5))
-            single_mask_inv_print = self.get_mask_inv(print_['image'])
+        gap = print_dict.get('gap', [[0, 0]])[0]
-            dim_max = max(painting_dict['dim_image_h'], painting_dict['dim_image_w'])
+        painting_dict['tile_print'], painting_dict['mask_inv_print'] = tile_image(pattern=print_['image'],
-            dim_pattern = (int(dim_max * print_['scale'] / 5), int(dim_max * print_['scale'] / 5))
+                                                                                  mask=print_['mask'],
-            if not is_single:
+                                                                                  dim=dim_pattern,
-                self.random_seed = random.randint(0, 1000)
+                                                                                  gap_x=gap[0],
-                # 如果print 模式为overall 且 有角度的话 ， 组合的print为正方形，方便裁剪
+                                                                                  gap_y=gap[1],
-                if "print_angle_list" in print_dict.keys() and print_dict['print_angle_list'][0] != 0:
+                                                                                  canvas_h=painting_dict['dim_image_h'],
-                    painting_dict['mask_inv_print'] = self.tile_image(single_mask_inv_print, dim_pattern, print_['scale'], dim_max, dim_max, painting_dict['location'], trigger=True)
+                                                                                  canvas_w=painting_dict['dim_image_w'],
-                    painting_dict['tile_print'] = self.tile_image(print_['image'], dim_pattern, print_['scale'], dim_max, dim_max, painting_dict['location'], trigger=True)
+                                                                                  location=painting_dict['location'],
-                else:
+                                                                                  angle=int(print_.get('print_angle_list', [0])[0]))
-                    painting_dict['mask_inv_print'] = self.tile_image(single_mask_inv_print, dim_pattern, print_['scale'], painting_dict['dim_image_h'], painting_dict['dim_image_w'], painting_dict['location'], trigger=True)
-                    painting_dict['tile_print'] = self.tile_image(print_['image'], dim_pattern, print_['scale'], painting_dict['dim_image_h'], painting_dict['dim_image_w'], painting_dict['location'], trigger=True)
-            else:
-                painting_dict['mask_inv_print'] = self.tile_image(single_mask_inv_print, dim_pattern, print_['scale'], painting_dict['dim_image_h'], painting_dict['dim_image_w'], painting_dict['location'])
-                painting_dict['tile_print'] = self.tile_image(print_['image'], dim_pattern, print_['scale'], painting_dict['dim_image_h'], painting_dict['dim_image_w'], painting_dict['location'])
-                painting_dict['dim_print_h'], painting_dict['dim_print_w'] = dim_pattern
         return painting_dict
 
     def tile_image(self, pattern, dim, scale, dim_image_h, dim_image_w, location, trigger=False):
@@ -374,51 +266,37 @@ class PrintPainting:
             mask_inv = cv2.inRange(print_tile, lower, upper)
             return mask_inv
         else:
-            # bg_color = cv2.cvtColor(print_, cv2.COLOR_BGR2LAB)[0][0]
-            # print_tile = cv2.cvtColor(print_, cv2.COLOR_BGR2LAB)
-            # bg_l, bg_a, bg_b = bg_color[0], bg_color[1], bg_color[2]
-            # bg_L_high, bg_L_low = self.get_low_high_lab(bg_l, L=True)
-            # bg_a_high, bg_a_low = self.get_low_high_lab(bg_a)
-            # bg_b_high, bg_b_low = self.get_low_high_lab(bg_b)
-            # lower = np.array([bg_L_low, bg_a_low, bg_b_low])
-            # upper = np.array([bg_L_high, bg_a_high, bg_b_high])
-
-            # print_tile = cv2.cvtColor(print_, cv2.COLOR_BGR2LAB)
-            # mask_inv = cv2.cvtColor(print_tile, cv2.COLOR_BGR2GRAY)
-
-            # mask_inv = cv2.cvtColor(print_, cv2.COLOR_BGR2GRAY)
             mask_inv = np.zeros(print_.shape[:2], dtype=np.uint8)
             return mask_inv
 
     @staticmethod
     def printpaint(result, painting_dict, print_=False):
-
+        if print_:
-        if print_ and painting_dict['Trigger']:
             print_mask = cv2.bitwise_and(result['mask'], cv2.bitwise_not(painting_dict['mask_inv_print']))
             img_fg = cv2.bitwise_and(painting_dict['tile_print'], painting_dict['tile_print'], mask=print_mask)
         else:
             print_mask = result['mask']
             img_fg = result['final_image']
-        if print_ and not painting_dict['Trigger']:
+        # if print_ and not painting_dict['Trigger']:
-            index_ = None
+        #     index_ = None
-            try:
+        #     try:
-                index_ = len(painting_dict['location'])
+        #         index_ = len(painting_dict['location'])
-            except:
+        #     except:
-                assert f'there must be parameter of location if choose IfSingle'
+        #         assert f'there must be parameter of location if choose IfSingle'
-
+        #
-            for i in range(index_):
+        #     for i in range(index_):
-                start_h, start_w = int(painting_dict['location'][i][1]), int(painting_dict['location'][i][0])
+        #         start_h, start_w = int(painting_dict['location'][i][1]), int(painting_dict['location'][i][0])
-
+        #
-                length_h = min(start_h + painting_dict['dim_print_h'], img_fg.shape[0])
+        #         length_h = min(start_h + painting_dict['dim_print_h'], img_fg.shape[0])
-                length_w = min(start_w + painting_dict['dim_print_w'], img_fg.shape[1])
+        #         length_w = min(start_w + painting_dict['dim_print_w'], img_fg.shape[1])
-
+        #
-                change_region = img_fg[start_h: length_h, start_w: length_w, :]
+        #         change_region = img_fg[start_h: length_h, start_w: length_w, :]
-                # problem in change_mask
+        #         # problem in change_mask
-                change_mask = print_mask[start_h: length_h, start_w: length_w]
+        #         change_mask = print_mask[start_h: length_h, start_w: length_w]
-                # get real part into change mask
+        #         # get real part into change mask
-                _, change_mask = cv2.threshold(change_mask, 220, 255, cv2.THRESH_BINARY)
+        #         _, change_mask = cv2.threshold(change_mask, 220, 255, cv2.THRESH_BINARY)
-                cv2.bitwise_not(painting_dict['mask_inv_print'])
+        #         cv2.bitwise_not(painting_dict['mask_inv_print'])
-                img_fg[start_h:start_h + painting_dict['dim_print_h'], start_w:start_w + painting_dict['dim_print_w'], :] = change_region
+        #         img_fg[start_h:start_h + painting_dict['dim_print_h'], start_w:start_w + painting_dict['dim_print_w'], :] = change_region
 
         clothes_mask_print = cv2.bitwise_not(print_mask)
 
@@ -440,21 +318,21 @@ class PrintPainting:
         image = oss_get_image(oss_client=self.minio_client, bucket=bucket_name, object_name=object_name, data_type="PIL")
         # 判断图片格式，如果是RGBA 则贴在一张纯白图片上 防止透明转黑
         if image.mode == "RGBA":
+            mask_pil = image.split()[3]
             new_background = Image.new('RGB', image.size, (255, 255, 255))
             new_background.paste(image, mask=image.split()[3])
             image = new_background
+        else:
+            mask_pil = Image.new('L', image.size, 255)  # L=灰度图，255=纯白
         print_dict['image'] = cv2.cvtColor(np.asarray(image), cv2.COLOR_RGB2BGR)
+        print_dict['mask'] = cv2.threshold(np.array(mask_pil), 127, 255, cv2.THRESH_BINARY)[1]
+
         return print_dict
 
     def crop_image(self, image, image_size_h, image_size_w, location, print_shape):
         print_w = print_shape[1]
         print_h = print_shape[0]
 
-        random.seed(self.random_seed)
-        # logging.info(f'overall print location : {location}')
-        # x_offset = random.randint(0, image.shape[0] - image_size_h)
-        # y_offset = random.randint(0, image.shape[1] - image_size_w)
-
         # 1.拿到偏移量后和resize后的print宽高取余 得到真正偏移量
         # 偏移量增加2分之print.w 使坐标位于图中间  如果要位于左上角删除+ print_w // 2 即可
         x_offset = print_w - int(location[0][1] % print_w) + print_w // 2
@@ -596,3 +474,118 @@ class PrintPainting:
             cropped_img = resized_img[start_y:start_y + target_height, :]
 
         return cropped_img
+
+
+def tile_image(pattern, mask, dim, gap_x, gap_y, canvas_h, canvas_w, location, angle=0):
+    """
+    按照指定的 X/Y 间距平铺印花，并支持旋转
+    【修改版】以被平铺图案的【中心】作为平铺基准点
+
+    :param location: [[center_y, center_x]]  → 第一个图案中心的坐标
+    :param angle: 旋转角度 (度数, 逆时针)
+    """
+    # 1. 确保输入是 RGBA
+    if pattern.shape[2] == 3:
+        pattern = cv2.cvtColor(pattern, cv2.COLOR_BGR2BGRA)
+
+    # 2. 缩放与旋转印花
+    resized_p = cv2.resize(pattern, dim, interpolation=cv2.INTER_AREA)
+    rotated_p = rotate_image(resized_p, angle)
+    p_h, p_w = rotated_p.shape[:2]
+
+    # 3. 创建透明单元格（图案放在单元格中心）
+    cell_h = p_h + gap_y
+    cell_w = p_w + gap_x
+
+    unit_cell = np.zeros((cell_h, cell_w, 4), dtype=np.uint8)
+
+    # 计算图案在单元格中的左上角位置（让图案居中）
+    start_y = (cell_h - p_h) // 2
+    start_x = (cell_w - p_w) // 2
+    unit_cell[start_y:start_y + p_h, start_x:start_x + p_w, :] = rotated_p
+
+    # 4. 执行平铺
+    tiles_y = (canvas_h // cell_h) + 3  # 多加一点余量更安全
+    tiles_x = (canvas_w // cell_w) + 3
+    full_tiled = np.tile(unit_cell, (tiles_y, tiles_x, 1))
+
+    # 5. 计算偏移（关键修改：以中心为基准）
+    center_y, center_x = location[0][0], location[0][1]  # 第一个图案的中心位置
+
+    # 计算从哪个位置开始裁剪，才能让中心落在指定坐标
+    offset_y = int((center_y - (p_h // 2)) % cell_h)
+    offset_x = int((center_x - (p_w // 2)) % cell_w)
+
+    tiled_layer = full_tiled[offset_y: offset_y + canvas_h,
+    offset_x: offset_x + canvas_w]
+
+    # 6. 创建纯白色背景并合成（保持你原来的风格）
+    white_background = np.full((canvas_h, canvas_w, 3), 255, dtype=np.uint8)
+
+    tiled_bgr = tiled_layer[:, :, :3]
+    alpha_mask = tiled_layer[:, :, 3] / 255.0
+    alpha_mask = cv2.merge([alpha_mask, alpha_mask, alpha_mask])
+
+    tiled_print = (tiled_bgr * alpha_mask + white_background * (1 - alpha_mask)).astype(np.uint8)
+
+    # ====================== 处理 Mask ======================
+    # Mask 也同样居中处理
+    resized_mask = cv2.resize(mask, dim, interpolation=cv2.INTER_NEAREST)
+    rotated_mask = rotate_image(resized_mask, angle)  # 注意：mask也需要旋转
+
+    unit_mask = np.zeros((cell_h, cell_w), dtype=np.uint8)
+    unit_mask[start_y:start_y + p_h, start_x:start_x + p_w] = rotated_mask
+
+    full_mask_tiled = np.tile(unit_mask, (tiles_y, tiles_x))
+    tiled_mask = full_mask_tiled[offset_y: offset_y + canvas_h,
+    offset_x: offset_x + canvas_w]
+
+    return tiled_print, cv2.bitwise_not(tiled_mask)
+
+
+def rotate_image(image, angle):
+    """
+    旋转图片并保持完整内容（自动扩大画布）
+    """
+    if angle == 0:
+        return image
+
+    (h, w) = image.shape[:2]
+    (cX, cY) = (w // 2, h // 2)
+
+    # 获取旋转矩阵
+    M = cv2.getRotationMatrix2D((cX, cY), angle, 1.0)
+
+    # 计算旋转后新边界的 sine 和 cosine
+    cos = np.abs(M[0, 0])
+    sin = np.abs(M[0, 1])
+
+    # 计算新的画布尺寸
+    nW = int((h * sin) + (w * cos))
+    nH = int((h * cos) + (w * sin))
+
+    # 调整旋转矩阵以考虑平移
+    M[0, 2] += (nW / 2) - cX
+    M[1, 2] += (nH / 2) - cY
+
+    # 执行旋转
+    return cv2.warpAffine(image, M, (nW, nH))
+
+
+def crop_image(image, image_size_h, image_size_w, location, print_shape):
+    print_w = print_shape[1]
+    print_h = print_shape[0]
+
+    # 1.拿到偏移量后和resize后的print宽高取余 得到真正偏移量
+    # 偏移量增加2分之print.w 使坐标位于图中间  如果要位于左上角删除+ print_w // 2 即可
+    x_offset = print_w - int(location[0][1] % print_w) + print_w // 2
+    y_offset = print_h - int(location[0][0] % print_h) + print_h // 2
+
+    # y_offset = int(location[0][0])
+    # x_offset = int(location[0][1])
+
+    if len(image.shape) == 2:
+        image = image[x_offset: x_offset + image_size_h, y_offset: y_offset + image_size_w]
+    elif len(image.shape) == 3:
+        image = image[x_offset: x_offset + image_size_h, y_offset: y_offset + image_size_w, :]
+    return image

app/service/design_fast/pipeline/segmentation.py

@@ -34,22 +34,25 @@ class Segmentation:
             result['mask'] = result['front_mask'] + result['back_mask']
         else:
             # preview 过模型 不缓存
-            if "preview_submit" in result.keys() and result['preview_submit'] == "preview":
+            if result.get("design_type", None) == "merge":
-                # 推理获得seg 结果
                 seg_result = get_seg_result(result['image'])
-            # submit 过模型 缓存
+            # 默认design 模式 - 过模型 缓存
-            elif "preview_submit" in result.keys() and result['preview_submit'] == "submit":
+            # elif result.get("design_type", None) == "submit":
-                # 推理获得seg 结果
+            # 推理获得seg 结果
-                seg_result = get_seg_result(result['image'])
+            # seg_result = get_seg_result(result['image'])
-                self.save_seg_result(seg_result, result['image_id'])
+            # self.save_seg_result(seg_result, result['image_id'])
-            #  null  正常流程 加载本地缓存 无缓存则过模型
+
+            # 默认模式- 加载模型,找不到则过模型推理,推理后保存到本地
             else:
                 # 本地查询seg 缓存是否存在
                 _, seg_result = self.load_seg_result(result["image_id"])
                 # 判断缓存和实际图片size是否相同
+                _ = False
                 if not _ or result["image"].shape[:2] != seg_result.shape:
                     # 推理获得seg 结果
                     seg_result = get_seg_result(result['image'])
+                    if result['name'] == 'others':
+                        seg_result = seg_result.clip(max=1)
                     self.save_seg_result(seg_result, result['image_id'])
             result['seg_result'] = seg_result
 

app/service/design_fast/pipeline/split.py

@@ -4,6 +4,7 @@ import logging
 import cv2
 import numpy as np
 from PIL import Image
+from celery.bin.result import result
 
 from app.service.design_fast.utils.conversion_image import rgb_to_rgba
 from app.service.design_fast.utils.transparent import sketch_to_transparent
@@ -19,105 +20,122 @@ class Split(object):
     def __call__(self, result):
         try:
             if result['name'] in ('outwear', 'dress', 'blouse', 'skirt', 'trousers', 'tops', 'bottoms', 'others'):
-                ori_front_mask = result['front_mask'].copy()
+                if result.get('design_type', None) == 'merge':
-                ori_back_mask = result['back_mask'].copy()
+                    ori_front_mask = result['front_mask'].copy()
+                    ori_back_mask = result['back_mask'].copy()
 
-                if result['resize_scale'][0] == 1.0 and result['resize_scale'][1] == 1.0:
+                    if result['resize_scale'][0] == 1.0 and result['resize_scale'][1] == 1.0:
-                    front_mask = result['front_mask']
+                        front_mask = result['front_mask']
-                    back_mask = result['back_mask']
+                        back_mask = result['back_mask']
-                else:
-                    height, width = result['front_mask'].shape[:2]
-                    new_width = int(width * result['resize_scale'][0])
-                    new_height = int(height * result['resize_scale'][1])
-
-                    front_mask = cv2.resize(result['front_mask'], (new_width, new_height), interpolation=cv2.INTER_AREA)
-                    back_mask = cv2.resize(result['back_mask'], (new_width, new_height), interpolation=cv2.INTER_AREA)
-
-                rgba_image = rgb_to_rgba(result['final_image'], front_mask + back_mask)
-                new_size = (int(rgba_image.shape[1] * result["scale"]), int(rgba_image.shape[0] * result["scale"]))
-                rgba_image = cv2.resize(rgba_image, new_size, interpolation=cv2.INTER_AREA)
-                result_front_image = np.zeros_like(rgba_image)
-                front_mask = cv2.resize(front_mask, new_size, interpolation=cv2.INTER_AREA)
-                result_front_image[front_mask != 0] = rgba_image[front_mask != 0]
-                result_front_image_pil = Image.fromarray(cv2.cvtColor(result_front_image, cv2.COLOR_BGR2RGBA))
-                if 'transparent' in result.keys():
-                    #  用户自选区域transparent
-                    transparent = result['transparent']
-                    if transparent['mask_url'] is not None and transparent['mask_url'] != "":
-                        # 预处理用户自选区mask
-                        seg_mask = oss_get_image(oss_client=self.minio_client, bucket=transparent['mask_url'].split('/')[0], object_name=transparent['mask_url'][transparent['mask_url'].find('/') + 1:], data_type="cv2")
-                        seg_mask = cv2.resize(seg_mask, new_size, interpolation=cv2.INTER_AREA)
-                        # 转换颜色空间为 RGB（OpenCV 默认是 BGR）
-                        image_rgb = cv2.cvtColor(seg_mask, cv2.COLOR_BGR2RGB)
-
-                        r, g, b = cv2.split(image_rgb)
-                        blue_mask = b > r
-
-                        # 创建红色和绿色掩码
-                        transparent_mask = np.array(blue_mask, dtype=np.uint8) * 255
-                        result_front_image_pil = sketch_to_transparent(result_front_image_pil, transparent_mask, transparent["scale"])
                     else:
-                        result_front_image_pil = sketch_to_transparent(result_front_image_pil, front_mask, transparent["scale"])
+                        height, width = result['front_mask'].shape[:2]
-                result['front_image'], result["front_image_url"], _ = upload_png_mask(self.minio_client, result_front_image_pil, f'{generate_uuid()}', mask=None)
+                        new_width = int(width * result['resize_scale'][0])
+                        new_height = int(height * result['resize_scale'][1])
 
-                # 前片部分 (红图部分)
+                        front_mask = cv2.resize(result['front_mask'], (new_width, new_height), interpolation=cv2.INTER_AREA)
-                # height, width = front_mask.shape
+                        back_mask = cv2.resize(result['back_mask'], (new_width, new_height), interpolation=cv2.INTER_AREA)
-                # mask_image = np.zeros((height, width, 3))
+                        result['merge_image'] = cv2.resize(result['merge_image'], (new_width, new_height), interpolation=cv2.INTER_AREA)
-                # mask_image[front_mask != 0] = [0, 0, 255]
 
-                # 切换为原始图片尺寸-------------------------------
+                    rgba_image = rgb_to_rgba(result['merge_image'], front_mask + back_mask)
-                height, width = ori_front_mask.shape
+                    new_size = (int(rgba_image.shape[1] * result["scale"]), int(rgba_image.shape[0] * result["scale"]))
-                mask_image = np.zeros((height, width, 3))
+                    rgba_image = cv2.resize(rgba_image, new_size, interpolation=cv2.INTER_AREA)
-                mask_image[ori_front_mask != 0] = [0, 0, 255]
+                    result_front_image = np.zeros_like(rgba_image)
-                # -----------------------------------------------
+                    front_mask = cv2.resize(front_mask, new_size, interpolation=cv2.INTER_AREA)
+                    result_front_image[front_mask != 0] = rgba_image[front_mask != 0]
+                    result_front_image_pil = Image.fromarray(cv2.cvtColor(result_front_image, cv2.COLOR_BGR2RGBA))
+                    result['front_image'], result["front_image_url"], _ = upload_png_mask(self.minio_client, result_front_image_pil, f'{generate_uuid()}', mask=None)
 
-                # if result["name"] in ('blouse', 'dress', 'outwear', 'tops'):
+                    height, width = ori_front_mask.shape
-                #     result_back_image = np.zeros_like(rgba_image)
+                    mask_image = np.zeros((height, width, 3))
-                #     back_mask = cv2.resize(back_mask, new_size, interpolation=cv2.INTER_AREA)
+                    mask_image[ori_front_mask != 0] = [0, 0, 255]
-                #     result_back_image[back_mask != 0] = rgba_image[back_mask != 0]
+                    mask_image[ori_back_mask != 0] = [0, 255, 0]
-                #     result_back_image_pil = Image.fromarray(cvtColor(result_back_image, COLOR_BGR2RGBA))
+                    rbga_mask = rgb_to_rgba(mask_image, ori_front_mask + ori_back_mask)
-                #     result['back_image'], result["back_image_url"], _ = upload_png_mask(self.minio_client, result_back_image_pil, f'{generate_uuid()}', mask=None)
+                    mask_pil = Image.fromarray(cv2.cvtColor(rbga_mask.astype(np.uint8), cv2.COLOR_BGR2RGBA))
-                #     mask_image[back_mask != 0] = [0, 255, 0]
+                    image_data = io.BytesIO()
-                #
+                    mask_pil.save(image_data, format='PNG')
-                #     rbga_mask = rgb_to_rgba(mask_image, front_mask + back_mask)
+                    image_data.seek(0)
-                #     mask_pil = Image.fromarray(cvtColor(rbga_mask.astype(np.uint8), COLOR_BGR2RGBA))
+                    image_bytes = image_data.read()
-                #     image_data = io.BytesIO()
+                    req = oss_upload_image(oss_client=self.minio_client, bucket="aida-clothing", object_name=f"mask/mask_{generate_uuid()}.png", image_bytes=image_bytes)
-                #     mask_pil.save(image_data, format='PNG')
+                    result['mask_url'] = req.bucket_name + "/" + req.object_name
-                #     image_data.seek(0)
-                #     image_bytes = image_data.read()
-                #     req = oss_upload_image(oss_client=self.minio_client, bucket=AIDA_CLOTHING, object_name=f"mask/mask_{generate_uuid()}.png", image_bytes=image_bytes)
-                #     result['mask_url'] = req.bucket_name + "/" + req.object_name
-                # else:
-                #     rbga_mask = rgb_to_rgba(mask_image, front_mask)
-                #     mask_pil = Image.fromarray(cvtColor(rbga_mask.astype(np.uint8), COLOR_BGR2RGBA))
-                #     image_data = io.BytesIO()
-                #     mask_pil.save(image_data, format='PNG')
-                #     image_data.seek(0)
-                #     image_bytes = image_data.read()
-                #     req = oss_upload_image(oss_client=self.minio_client, bucket=AIDA_CLOTHING, object_name=f"mask/mask_{generate_uuid()}.png", image_bytes=image_bytes)
-                #     result['mask_url'] = req.bucket_name + "/" + req.object_name
-                #     result['back_image'] = None
-                #     result["back_image_url"] = None
-                #     # result["back_mask_url"] = None
-                #     # result['back_mask_image'] = None
 
-                result_back_image = np.zeros_like(rgba_image)
-                back_mask = cv2.resize(back_mask, new_size, interpolation=cv2.INTER_AREA)
-                result_back_image[back_mask != 0] = rgba_image[back_mask != 0]
-                result_back_image_pil = Image.fromarray(cv2.cvtColor(result_back_image, cv2.COLOR_BGR2RGBA))
-                result['back_image'], result["back_image_url"], _ = upload_png_mask(self.minio_client, result_back_image_pil, f'{generate_uuid()}', mask=None)
 
-                # mask_image[back_mask != 0] = [0, 255, 0]
+                    result_back_image = np.zeros_like(rgba_image)
-                mask_image[ori_back_mask != 0] = [0, 255, 0]
+                    back_mask = cv2.resize(back_mask, new_size, interpolation=cv2.INTER_AREA)
+                    result_back_image[back_mask != 0] = rgba_image[back_mask != 0]
+                    result_back_image_pil = Image.fromarray(cv2.cvtColor(result_back_image, cv2.COLOR_BGR2RGBA))
+                    result['back_image'], result["back_image_url"], _ = upload_png_mask(self.minio_client, result_back_image_pil, f'{generate_uuid()}', mask=None)
+                    return result
+                else:
+                    ori_front_mask = result['front_mask'].copy()
+                    ori_back_mask = result['back_mask'].copy()
 
-                rbga_mask = rgb_to_rgba(mask_image, ori_front_mask + ori_back_mask)
+                    if result['resize_scale'][0] == 1.0 and result['resize_scale'][1] == 1.0:
-                mask_pil = Image.fromarray(cv2.cvtColor(rbga_mask.astype(np.uint8), cv2.COLOR_BGR2RGBA))
+                        front_mask = result['front_mask']
-                image_data = io.BytesIO()
+                        back_mask = result['back_mask']
-                mask_pil.save(image_data, format='PNG')
+                    else:
-                image_data.seek(0)
+                        height, width = result['front_mask'].shape[:2]
-                image_bytes = image_data.read()
+                        new_width = int(width * result['resize_scale'][0])
-                req = oss_upload_image(oss_client=self.minio_client, bucket="aida-clothing", object_name=f"mask/mask_{generate_uuid()}.png", image_bytes=image_bytes)
+                        new_height = int(height * result['resize_scale'][1])
-                result['mask_url'] = req.bucket_name + "/" + req.object_name
+
+                        front_mask = cv2.resize(result['front_mask'], (new_width, new_height), interpolation=cv2.INTER_AREA)
+                        back_mask = cv2.resize(result['back_mask'], (new_width, new_height), interpolation=cv2.INTER_AREA)
+
+                    rgba_image = rgb_to_rgba(result['final_image'], front_mask + back_mask)
+                    new_size = (int(rgba_image.shape[1] * result["scale"]), int(rgba_image.shape[0] * result["scale"]))
+                    rgba_image = cv2.resize(rgba_image, new_size, interpolation=cv2.INTER_AREA)
+                    result_front_image = np.zeros_like(rgba_image)
+                    front_mask = cv2.resize(front_mask, new_size, interpolation=cv2.INTER_AREA)
+                    result_front_image[front_mask != 0] = rgba_image[front_mask != 0]
+                    result_front_image_pil = Image.fromarray(cv2.cvtColor(result_front_image, cv2.COLOR_BGR2RGBA))
+                    if 'transparent' in result.keys():
+                        #  用户自选区域transparent
+                        transparent = result['transparent']
+                        if transparent['mask_url'] is not None and transparent['mask_url'] != "":
+                            # 预处理用户自选区mask
+                            seg_mask = oss_get_image(oss_client=self.minio_client, bucket=transparent['mask_url'].split('/')[0], object_name=transparent['mask_url'][transparent['mask_url'].find('/') + 1:], data_type="cv2")
+                            seg_mask = cv2.resize(seg_mask, new_size, interpolation=cv2.INTER_AREA)
+                            # 转换颜色空间为 RGB（OpenCV 默认是 BGR）
+                            image_rgb = cv2.cvtColor(seg_mask, cv2.COLOR_BGR2RGB)
+
+                            r, g, b = cv2.split(image_rgb)
+                            blue_mask = b > r
+
+                            # 创建红色和绿色掩码
+                            transparent_mask = np.array(blue_mask, dtype=np.uint8) * 255
+                            result_front_image_pil = sketch_to_transparent(result_front_image_pil, transparent_mask, transparent["scale"])
+                        else:
+                            result_front_image_pil = sketch_to_transparent(result_front_image_pil, front_mask, transparent["scale"])
+                    result['front_image'], result["front_image_url"], _ = upload_png_mask(self.minio_client, result_front_image_pil, f'{generate_uuid()}', mask=None)
+
+                    height, width = ori_front_mask.shape
+                    mask_image = np.zeros((height, width, 3))
+                    mask_image[ori_front_mask != 0] = [0, 0, 255]
+
+                    result_back_image = np.zeros_like(rgba_image)
+                    back_mask = cv2.resize(back_mask, new_size, interpolation=cv2.INTER_AREA)
+                    result_back_image[back_mask != 0] = rgba_image[back_mask != 0]
+                    result_back_image_pil = Image.fromarray(cv2.cvtColor(result_back_image, cv2.COLOR_BGR2RGBA))
+                    result['back_image'], result["back_image_url"], _ = upload_png_mask(self.minio_client, result_back_image_pil, f'{generate_uuid()}', mask=None)
+
+                    # mask_image[back_mask != 0] = [0, 255, 0]
+                    mask_image[ori_back_mask != 0] = [0, 255, 0]
+
+                    rbga_mask = rgb_to_rgba(mask_image, ori_front_mask + ori_back_mask)
+                    mask_pil = Image.fromarray(cv2.cvtColor(rbga_mask.astype(np.uint8), cv2.COLOR_BGR2RGBA))
+                    image_data = io.BytesIO()
+                    mask_pil.save(image_data, format='PNG')
+                    image_data.seek(0)
+                    image_bytes = image_data.read()
+                    req = oss_upload_image(oss_client=self.minio_client, bucket="aida-clothing", object_name=f"mask/mask_{generate_uuid()}.png", image_bytes=image_bytes)
+                    result['mask_url'] = req.bucket_name + "/" + req.object_name
+
+                    # 创建中间图层(未分割图层) 1.color + overall_print  2.color + overall_print + print
+                    result_pattern_overall_image_pil = Image.fromarray(cv2.cvtColor(rgb_to_rgba(result['no_seg_sketch_overall'], ori_front_mask + ori_back_mask), cv2.COLOR_BGR2RGBA))
+                    result['pattern_overall_image'], result['pattern_overall_image_url'], _ = upload_png_mask(self.minio_client, result_pattern_overall_image_pil, f'{generate_uuid()}')
+
+                    result_pattern_print_image_pil = Image.fromarray(cv2.cvtColor(rgb_to_rgba(result['no_seg_sketch_print'], ori_front_mask + ori_back_mask), cv2.COLOR_BGR2RGBA))
+                    result['pattern_print_image'], result['pattern_print_image_url'], _ = upload_png_mask(self.minio_client, result_pattern_print_image_pil, f'{generate_uuid()}')
+                    return result
             else:
                 ori_front_mask, ori_back_mask = None, None
             # 创建中间图层(未分割图层) 1.color + overall_print  2.color + overall_print + print
@@ -127,5 +145,6 @@ class Split(object):
             result_pattern_print_image_pil = Image.fromarray(cv2.cvtColor(rgb_to_rgba(result['no_seg_sketch_print'], ori_front_mask + ori_back_mask), cv2.COLOR_BGR2RGBA))
             result['pattern_print_image'], result['pattern_print_image_url'], _ = upload_png_mask(self.minio_client, result_pattern_print_image_pil, f'{generate_uuid()}')
             return result
+
         except Exception as e:
             logging.warning(f"split runtime exception : {e} image_id : {result['image_id']}")

app/service/design_fast/utils/design_ensemble.py

@@ -10,12 +10,12 @@
 import logging
 
 import cv2
-import mmcv
 import numpy as np
 import torch
 import tritonclient.http as httpclient
 
 from app.core.config import DESIGN_MODEL_URL, DESIGN_MODEL_NAME
+from app.service.utils.image_normalize import my_imnormalize
 
 """
     keypoint
@@ -24,14 +24,14 @@ from app.core.config import DESIGN_MODEL_URL, DESIGN_MODEL_NAME
 
 
 def keypoint_preprocess(img_path):
-    img = mmcv.imread(img_path)
+    img = img_path
     img = cv2.copyMakeBorder(img, 25, 25, 25, 25, cv2.BORDER_CONSTANT, value=[255, 255, 255])
     img_scale = (256, 256)
     h, w = img.shape[:2]
     img = cv2.resize(img, img_scale)
     w_scale = img_scale[0] / w
     h_scale = img_scale[1] / h
-    img = mmcv.imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
+    img = my_imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
     preprocessed_img = np.expand_dims(img.transpose(2, 0, 1), axis=0)
     return preprocessed_img, (w_scale, h_scale)
 
@@ -78,7 +78,7 @@ def keypoint_postprocess(output, scale_factor):
 
 # KNet
 def seg_preprocess(img_path):
-    img = mmcv.imread(img_path)
+    img = img_path
     ori_shape = img.shape[:2]
     img_scale_w, img_scale_h = ori_shape
     if ori_shape[0] > 1024:
@@ -87,12 +87,12 @@ def seg_preprocess(img_path):
         img_scale_h = 1024
     # 如果图片size任意一边 大于 1024， 则会resize 成1024
     if ori_shape != (img_scale_w, img_scale_h):
-        # mmcv.imresize(img, img_scale_h, img_scale_w) # 老代码 引以为戒！哈哈哈~ h和w写反了
+        # my_imnormalize(img, img_scale_h, img_scale_w) # 老代码 引以为戒！哈哈哈~ h和w写反了
         img = cv2.resize(img, (img_scale_h, img_scale_w))
 
     # 扩充25的白边
     img = cv2.copyMakeBorder(img, 25, 25, 25, 25, cv2.BORDER_CONSTANT, value=[255, 255, 255])
-    img = mmcv.imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
+    img = my_imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
     preprocessed_img = np.expand_dims(img.transpose(2, 0, 1), axis=0)
     return preprocessed_img, ori_shape
 

app/service/design_fast/utils/organize.py

@@ -23,20 +23,23 @@ def organize_clothing(layer):
     front_layer = dict(priority=layer['priority'] if layer.get("layer_order", False) else PRIORITY_DICT.get(f'{layer["name"].lower()}_front', None),
                        name=f'{layer["name"].lower()}_front',
                        image=layer["front_image"],
+                       merge_image=layer["front_image"],
                        # mask_image=layer['front_mask_image'],
                        image_url=layer['front_image_url'],
-                       mask_url=layer['mask_url'],
+                       mask_url=layer.get("mask_url", None),
                        sacle=layer['scale'],
                        clothes_keypoint=layer['clothes_keypoint'],
                        position=start_point,
                        resize_scale=layer["resize_scale"],
                        mask=cv2.resize(layer['mask'], layer["front_image"].size),
                        gradient_string=layer['gradient_string'] if 'gradient_string' in layer.keys() else "",
-                       pattern_overall_image_url=layer['pattern_overall_image_url'],
+                       pattern_overall_image_url=layer.get('pattern_overall_image_url', None),
-                       pattern_print_image_url=layer['pattern_print_image_url'],
+                       pattern_print_image_url=layer.get('pattern_print_image_url', None),
 
-                       pattern_image=layer['pattern_image'],
+                       pattern_image=layer.get('pattern_image', None),
                        # back_perspective_url=layer['back_perspective_url'] if 'back_perspective_url' in layer.keys() else ""
+                       transpose=layer.get("transpose", [1, 1]),  # 默认为1, 1代表不镜像
+                       rotate=layer.get('rotate', 0),
                        )
     # 后片数据
     back_layer = dict(priority=-layer.get("priority", 0) if layer.get("layer_order", False) else PRIORITY_DICT.get(f'{layer["name"].lower()}_back', None),
@@ -44,16 +47,18 @@ def organize_clothing(layer):
                       image=layer["back_image"],
                       # mask_image=layer['back_mask_image'],
                       image_url=layer['back_image_url'],
-                      mask_url=layer['mask_url'],
+                      mask_url=layer.get('mask_url', None),
                       sacle=layer['scale'],
                       clothes_keypoint=layer['clothes_keypoint'],
                       position=start_point,
                       resize_scale=layer["resize_scale"],
                       mask=cv2.resize(layer['mask'], layer["front_image"].size),
                       gradient_string=layer['gradient_string'] if 'gradient_string' in layer.keys() else "",
-                      pattern_overall_image_url=layer['pattern_overall_image_url'],
+                      pattern_overall_image_url=layer.get('pattern_overall_image_url', None),
-                      pattern_print_image_url=layer['pattern_print_image_url'],
+                      pattern_print_image_url=layer.get('pattern_print_image_url', None),
                       # back_perspective_url=layer['back_perspective_url'] if 'back_perspective_url' in layer.keys() else ""
+                      transpose=layer.get("transpose", [1, 1]),  # 默认为1, 1代表不镜像
+                      rotate=layer.get('rotate', 0),
                       )
     return front_layer, back_layer
 
@@ -76,17 +81,19 @@ def organize_others(layer):
                        image=layer["front_image"],
                        # mask_image=layer['front_mask_image'],
                        image_url=layer['front_image_url'],
-                       mask_url=layer['mask_url'],
+                       mask_url=layer.get('mask_url', None),
                        sacle=layer['scale'],
                        clothes_keypoint=(0, 0),
                        position=start_point,
                        resize_scale=layer["resize_scale"],
                        mask=cv2.resize(layer['mask'], layer["front_image"].size),
                        gradient_string=layer['gradient_string'] if 'gradient_string' in layer.keys() else "",
-                       pattern_overall_image_url=layer['pattern_overall_image_url'],
+                       pattern_overall_image_url=layer.get('pattern_overall_image_url', None),
-                       pattern_print_image_url=layer['pattern_print_image_url'],
+                       pattern_print_image_url=layer.get('pattern_print_image_url', None),
-                       pattern_image=layer['pattern_image'],
+                       pattern_image=layer.get('pattern_image', None),
                        # back_perspective_url=layer['back_perspective_url'] if 'back_perspective_url' in layer.keys() else ""
+                       transpose=layer.get("transpose", [1, 1]),  # 默认为1, 1代表不镜像
+                       rotate=layer.get('rotate', 0),
                        )
     # 后片数据
     back_layer = dict(priority=-layer.get("priority", 0) if layer.get("layer_order", False) else PRIORITY_DICT.get(f'{layer["name"].lower()}_back', None),
@@ -94,16 +101,18 @@ def organize_others(layer):
                       image=layer["back_image"],
                       # mask_image=layer['back_mask_image'],
                       image_url=layer['back_image_url'],
-                      mask_url=layer['mask_url'],
+                      mask_url=layer.get('mask_url', None),
                       sacle=layer['scale'],
                       clothes_keypoint=(0, 0),
                       position=start_point,
                       resize_scale=layer["resize_scale"],
                       mask=cv2.resize(layer['mask'], layer["front_image"].size),
                       gradient_string=layer['gradient_string'] if 'gradient_string' in layer.keys() else "",
-                      pattern_overall_image_url=layer['pattern_overall_image_url'],
+                      pattern_overall_image_url=layer.get('pattern_overall_image_url', None),
-                      pattern_print_image_url=layer['pattern_print_image_url'],
+                      pattern_print_image_url=layer.get('pattern_print_image_url', None),
                       # back_perspective_url=layer['back_perspective_url'] if 'back_perspective_url' in layer.keys() else ""
+                      transpose=layer.get("transpose", [1, 1]),  # 默认为1, 1代表不镜像
+                      rotate=layer.get('rotate', 0),
                       )
     return front_layer, back_layer
 

app/service/design_fast/utils/synthesis_item.py

@@ -151,9 +151,11 @@ def synthesis(data, size, basic_info):
             if layer['image'] is not None:
                 if layer['name'] != "body":
                     test_image = Image.new('RGBA', size, (0, 0, 0, 0))
-                    test_image.paste(layer['image'], (layer['adaptive_position'][1], layer['adaptive_position'][0]), layer['image'])
+                    paste_img, position = transpose_rotate(layer, layer['image'])
+                    test_image.paste(paste_img, position, paste_img)
                     mask_data = np.where(all_mask > 0, 255, 0).astype(np.uint8)
                     mask_alpha = Image.fromarray(mask_data)
+                    mask_alpha.paste(paste_img.getchannel('A'), position, paste_img.getchannel('A'))
                     cropped_image = Image.composite(test_image, Image.new("RGBA", test_image.size, (255, 255, 255, 0)), mask_alpha)
                     base_image.paste(test_image, (0, 0), cropped_image)  # test_image 已经按照坐标贴到最大宽值的图片上 坐着这里坐标为00
                 else:
@@ -185,6 +187,111 @@ def synthesis(data, size, basic_info):
         logging.warning(f"synthesis runtime exception : {e}")
 
 
+def merge(data, size, basic_info):
+    # out_of_bounds_control: 是否允许服装越界  True 允许 False 不允许  默认情况允许
+    out_of_bounds_control = basic_info.get('out_of_bounds_control', True)
+    # 创建底图
+    base_image = Image.new('RGBA', size, (0, 0, 0, 0))
+    try:
+        all_mask_shape = (size[1], size[0])
+        body_mask = None
+        for d in data:
+            if d['name'] == 'body' or d['name'] == 'mannequin':
+                # 创建一个新的宽高透明图像， 把模特贴上去获取mask
+                transparent_image = Image.new("RGBA", size, (0, 0, 0, 0))
+                transparent_image.paste(d['image'], (d['adaptive_position'][1], d['adaptive_position'][0]), d['image'])  # 此处可变数组会被paste篡改值，所以使用下标获取position
+                body_mask = np.array(transparent_image.split()[3])
+
+                # 根据新的坐标获取新的肩点
+                left_shoulder = [x + y for x, y in zip(basic_info['body_point_test']['shoulder_left'], [d['adaptive_position'][1], d['adaptive_position'][0]])]
+                right_shoulder = [x + y for x, y in zip(basic_info['body_point_test']['shoulder_right'], [d['adaptive_position'][1], d['adaptive_position'][0]])]
+                body_mask[:min(left_shoulder[1], right_shoulder[1]), left_shoulder[0]:right_shoulder[0]] = 255
+        _, binary_body_mask = cv2.threshold(body_mask, 127, 255, cv2.THRESH_BINARY)
+        top_outer_mask = np.array(binary_body_mask)
+        bottom_outer_mask = np.array(binary_body_mask)
+        others_outer_mask = np.array(binary_body_mask)
+
+        top = True
+        bottom = True
+        others = True
+        i = len(data)
+        while i:
+            i -= 1
+            if top and data[i]['name'] in ["blouse_front", "outwear_front", "dress_front", "tops_front"]:
+                if out_of_bounds_control:
+                    top = True
+                else:
+                    top = False
+                mask_shape = data[i]['mask'].shape
+                y_offset, x_offset = data[i]['adaptive_position']
+                # 初始化叠加区域的起始和结束位置
+                all_y_start, all_y_end, mask_y_start, mask_y_end = positioning(all_mask_shape=all_mask_shape[0], mask_shape=mask_shape[0], offset=y_offset)
+                all_x_start, all_x_end, mask_x_start, mask_x_end = positioning(all_mask_shape=all_mask_shape[1], mask_shape=mask_shape[1], offset=x_offset)
+                # 将叠加区域赋值为相应的像素值
+                _, sketch_mask = cv2.threshold(data[i]['mask'], 127, 255, cv2.THRESH_BINARY)
+                background = np.zeros_like(top_outer_mask)
+                background[all_y_start:all_y_end, all_x_start:all_x_end] = sketch_mask[mask_y_start:mask_y_end, mask_x_start:mask_x_end]
+                top_outer_mask = background + top_outer_mask
+            elif bottom and data[i]['name'] in ["trousers_front", "skirt_front", "bottoms_front", "dress_front"]:
+                # bottom = False
+                mask_shape = data[i]['mask'].shape
+                y_offset, x_offset = data[i]['adaptive_position']
+                # 初始化叠加区域的起始和结束位置
+                all_y_start, all_y_end, mask_y_start, mask_y_end = positioning(all_mask_shape=all_mask_shape[0], mask_shape=mask_shape[0], offset=y_offset)
+                all_x_start, all_x_end, mask_x_start, mask_x_end = positioning(all_mask_shape=all_mask_shape[1], mask_shape=mask_shape[1], offset=x_offset)
+                # 将叠加区域赋值为相应的像素值
+                _, sketch_mask = cv2.threshold(data[i]['mask'], 127, 255, cv2.THRESH_BINARY)
+                background = np.zeros_like(top_outer_mask)
+                background[all_y_start:all_y_end, all_x_start:all_x_end] = sketch_mask[mask_y_start:mask_y_end, mask_x_start:mask_x_end]
+                bottom_outer_mask = background + bottom_outer_mask
+            elif others and data[i]['name'] in ['others_front']:
+                mask_shape = data[i]['mask'].shape
+                y_offset, x_offset = data[i]['adaptive_position']
+                # 初始化叠加区域的起始和结束位置
+                all_y_start, all_y_end, mask_y_start, mask_y_end = positioning(all_mask_shape=all_mask_shape[0], mask_shape=mask_shape[0], offset=y_offset)
+                all_x_start, all_x_end, mask_x_start, mask_x_end = positioning(all_mask_shape=all_mask_shape[1], mask_shape=mask_shape[1], offset=x_offset)
+                # 将叠加区域赋值为相应的像素值
+                _, sketch_mask = cv2.threshold(data[i]['mask'], 127, 255, cv2.THRESH_BINARY)
+                background = np.zeros_like(top_outer_mask)
+                background[all_y_start:all_y_end, all_x_start:all_x_end] = sketch_mask[mask_y_start:mask_y_end, mask_x_start:mask_x_end]
+                others_outer_mask = background + others_outer_mask
+                pass
+            elif bottom is False and top is False:
+                break
+
+        all_mask = cv2.bitwise_or(top_outer_mask, bottom_outer_mask)
+        all_mask = cv2.bitwise_or(all_mask, others_outer_mask)
+
+        for layer in data:
+            if layer['image'] is not None:
+                if layer['name'] != "body":
+                    test_image = Image.new('RGBA', size, (0, 0, 0, 0))
+                    paste_img, position = transpose_rotate(layer, layer['image'])
+                    test_image.paste(paste_img, position, paste_img)
+                    mask_data = np.where(all_mask > 0, 255, 0).astype(np.uint8)
+                    mask_alpha = Image.fromarray(mask_data)
+                    mask_alpha.paste(paste_img.getchannel('A'), position, paste_img.getchannel('A'))
+                    cropped_image = Image.composite(test_image, Image.new("RGBA", test_image.size, (255, 255, 255, 0)), mask_alpha)
+                    base_image.paste(test_image, (0, 0), cropped_image)  # test_image 已经按照坐标贴到最大宽值的图片上 坐着这里坐标为00
+                else:
+                    base_image.paste(layer['merge_image'], (layer['adaptive_position'][1], layer['adaptive_position'][0]), layer['merge_image'])
+
+        result_image = base_image
+
+        image_data = io.BytesIO()
+        result_image.save(image_data, format='PNG')
+        image_data.seek(0)
+
+        # oss upload
+        image_bytes = image_data.read()
+        bucket_name = "aida-results"
+        object_name = f'result_{generate_uuid()}.png'
+        oss_upload_image(oss_client=minio_client, bucket=bucket_name, object_name=object_name, image_bytes=image_bytes)
+        return f"{bucket_name}/{object_name}"
+    except Exception as e:
+        logging.warning(f"synthesis runtime exception : {e}")
+
+
 def synthesis_single(front_image, back_image):
     result_image = None
     if front_image:
@@ -232,3 +339,36 @@ def update_base_size_priority(layers):
     for info in layers:
         info['adaptive_position'] = (info['position'][0], info['position'][1] - min_x)
     return layers, (new_width, new_height)
+
+
+def transpose_rotate(layer, image):
+    # transpose[0]是左右 transpose[1]是上下
+    transpose = layer.get('transpose', [1, 1])  # 默认为1, 1代表不镜像
+
+    rotate = layer.get('rotate', 0)
+    paste_x, paste_y = layer['adaptive_position'][1], layer['adaptive_position'][0]
+    original_w = image.width
+    original_h = image.height
+    # transpose左右是1 上下是-1
+    if transpose[0] != 1:
+        # 左右
+        image = image.transpose(0)
+
+    if transpose[1] != 1:
+        # 上下
+        image = image.transpose(1)
+
+    if rotate:
+        image = image.rotate(-rotate, expand=True)
+        # 4. 计算粘贴位置以保持视觉中心一致
+        # 原本 (15, 36) 是 288*288 的左上角，我们计算其中心点
+        target_center_x = paste_x + original_w // 2
+        target_center_y = paste_y + original_h // 2
+
+        # 获取旋转后图像的新尺寸
+        new_w, new_h = image.size
+
+        # 计算新的左上角坐标，使得旋转后的图像中心依然在原定的中心位置
+        paste_x = target_center_x - new_w // 2
+        paste_y = target_center_y - new_h // 2
+    return image, (paste_x, paste_y)

app/service/design_pre_processing/service.py

@@ -7,7 +7,7 @@ import numpy as np
 import torch
 import tritonclient.grpc as grpcclient
 from minio import Minio
-from pymilvus import MilvusClient
+# from pymilvus import MilvusClient
 from urllib3.exceptions import ResponseError
 
 from app.core.config import settings, SR_MODEL_NAME, SR_TRITON_URL, MILVUS_TABLE_KEYPOINT, KEYPOINT_RESULT_TABLE_FIELD_SET
@@ -58,7 +58,21 @@ class DesignPreprocessing:
             if len(image.shape) == 2:
                 image = cv2.cvtColor(image, cv2.COLOR_GRAY2RGB)
             elif image.shape[2] == 4:  # 如果是四通道 mask
-                image = image[:, :, :3]
+                # 分离RGB和Alpha通道
+                bgr = image[:, :, :3]
+                alpha = image[:, :, 3]
+
+                # 创建白色背景（也可改为其他颜色，如(255,255,255)就是白色）
+                background_color = (255, 255, 255)
+                background = np.full_like(bgr, background_color)
+
+                # 将Alpha通道转换为掩码（0=透明，255=不透明）
+                alpha_mask = alpha / 255.0  # 归一化到0-1
+                alpha_mask = np.expand_dims(alpha_mask, axis=-1)  # 扩展维度，方便广播计算
+
+                # 混合背景和原图：透明区域显示背景色，不透明区域显示原图
+                image = (bgr * alpha_mask + background * (1 - alpha_mask)).astype(np.uint8)
+                # 此时image已经是3通道RGB，无需再执行image = image[:, :, :3]
             obj["image_obj"] = image
         return image_list
 
@@ -174,8 +188,9 @@ class DesignPreprocessing:
             scale = 0.4
             if waist_width / scale >= image_width:
                 add_width = int((waist_width / scale - image_width) / 2)
-                ret = cv2.copyMakeBorder(image['obj'], 0, 0, add_width, add_width, cv2.BORDER_CONSTANT, value=(256, 256, 256))
+                ret = cv2.copyMakeBorder(image['obj'], 0, 0, add_width, add_width, cv2.BORDER_CONSTANT, value=(255, 255, 255))
-                image_bytes = cv2.imencode(".jpg", ret)[1].tobytes()
+                img_rgba = cv2.cvtColor(ret, cv2.COLOR_RGB2RGBA)
+                image_bytes = cv2.imencode(".png", img_rgba)[1].tobytes()
                 # image['show_image_url'] = f"{image['image_url'].split('/', 1)[0]}/{self.minio_client.put_object(image['image_url'].split('/', 1)[0], image['image_url'].split('/', 1)[1].replace('.', '-show.'), io.BytesIO(image_bytes), len(image_bytes), content_type='image/jpeg').object_name}"
                 bucket_name = image['image_url'].split('/', 1)[0]
                 object_name = image['image_url'].split('/', 1)[1].replace('.', '-show.')
@@ -261,14 +276,15 @@ class DesignPreprocessing:
 
     def keypoint_cache(self, sketch):
         try:
-            client = MilvusClient(uri=settings.MILVUS_URL, token=settings.MILVUS_TOKEN, db_name=settings.MILVUS_ALIAS)
+            # client = MilvusClient(uri=settings.MILVUS_URL, token=settings.MILVUS_TOKEN, db_name=settings.MILVUS_ALIAS)
             keypoint_id = sketch['image_id']
-            res = client.query(
+            # res = client.query(
-                collection_name=MILVUS_TABLE_KEYPOINT,
+            #     collection_name=MILVUS_TABLE_KEYPOINT,
-                # ids=[keypoint_id],
+            #     # ids=[keypoint_id],
-                filter=f"keypoint_id == {keypoint_id}",
+            #     filter=f"keypoint_id == {keypoint_id}",
-                output_fields=['keypoint_vector', 'keypoint_site']
+            #     output_fields=['keypoint_vector', 'keypoint_site']
-            )
+            # )
+            res = []
             if len(res) == 0:
                 # 没有结果 直接推理拿结果 并保存
                 keypoint_infer_result = self.infer_keypoint_result(sketch)

app/service/generate_image/service_agent_tool_generate_image.py

@@ -11,7 +11,6 @@ import logging
 import uuid
 
 import cv2
-import mmcv
 import numpy as np
 import pandas as pd
 import torch
@@ -21,6 +20,7 @@ from minio import Minio
 from tritonclient.utils import np_to_triton_dtype
 
 from app.core.config import settings, FAST_GI_MODEL_URL, GI_MODEL_URL, DESIGN_MODEL_URL, FAST_GI_MODEL_NAME, GI_MODEL_NAME
+from app.service.utils.image_normalize import my_imnormalize
 from app.service.utils.new_oss_client import oss_upload_image
 
 logger = logging.getLogger()
@@ -86,10 +86,9 @@ class AgentToolGenerateImage:
 
     @staticmethod
     def preprocess(img):
-        img = mmcv.imread(img)
         img_scale = (224, 224)
         img = cv2.resize(img, img_scale)
-        img = mmcv.imnormalize(
+        img = my_imnormalize(
             img,
             mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]),
             to_rgb=True)

app/service/generate_image/service_generate_image.py

@@ -189,10 +189,10 @@ if __name__ == '__main__':
         tasks_id="123-89",
         prompt="a single item of sketch of dress, 4k, white background",
         image_url="aida-collection-element/89/Sketchboard/95f20cdc-e059-435c-b8b1-d04cc9e80c3d.png",
-        mode='img2img',
+        mode='txt2img',
         category="sketch",
         gender="Female",
-        version="fast"
+        version="hight"
     )
     server = GenerateImage(rd)
     print(server.get_result())

app/service/generate_image/utils/image_processing.py

@@ -2,23 +2,23 @@ import logging
 import time
 
 import cv2
-import mmcv
 import numpy as np
 import torch
 import tritonclient.http as httpclient
 
 from app.core.config import settings, DESIGN_MODEL_URL, DESIGN_MODEL_NAME
 from app.service.generate_image.utils.upload_sd_image import upload_stain_png_sd, upload_face_png_sd
+from app.service.utils.image_normalize import my_imnormalize
 
 logger = logging.getLogger()
 
 
 def seg_preprocess(img_path):
-    img = mmcv.imread(img_path)
+    img = img_path
     ori_shape = img.shape[:2]
     img_scale = ori_shape
     img = cv2.resize(img, img_scale)
-    img = mmcv.imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
+    img = my_imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
     preprocessed_img = np.expand_dims(img.transpose(2, 0, 1), axis=0)
     return preprocessed_img, ori_shape
 
@@ -242,10 +242,9 @@ def stain_detection(image, user_id, category, tasks_id, spot_size=100):
 
 def generate_category_recognition(image, gender):
     def preprocess(img):
-        img = mmcv.imread(img)
         img_scale = (224, 224)
         img = cv2.resize(img, img_scale)
-        img = mmcv.imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
+        img = my_imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
         preprocessed_img = np.expand_dims(img.transpose(2, 0, 1), axis=0)
         return preprocessed_img
 

app/service/lineart/service.py

@@ -1,7 +1,6 @@
 import logging
 
 import cv2
-import mmcv
 import numpy as np
 import torch
 import torch.nn.functional as F
@@ -10,6 +9,7 @@ from minio import Minio
 from app.core.config import settings
 from app.core.config import DESIGN_MODEL_URL
 from app.schemas.image2sketch import Image2SketchModel
+from app.service.utils.image_normalize import my_imnormalize
 from app.service.utils.new_oss_client import oss_get_image, oss_upload_image
 
 logger = logging.getLogger()
@@ -67,7 +67,7 @@ class LineArtService:
 
     @staticmethod
     def line_art_preprocess(image):
-        img = mmcv.imread(image)
+        img = image
         ori_shape = img.shape[:2]
         img_scale_w, img_scale_h = ori_shape
         if ori_shape[0] > 1024:
@@ -76,9 +76,9 @@ class LineArtService:
             img_scale_h = 1024
         # 如果图片size任意一边 大于 1024， 则会resize 成1024
         if ori_shape != (img_scale_w, img_scale_h):
-            # mmcv.imresize(img, img_scale_h, img_scale_w) # 老代码 引以为戒！哈哈哈~ h和w写反了
+            # my_imnormalize(img, img_scale_h, img_scale_w) # 老代码 引以为戒！哈哈哈~ h和w写反了
             img = cv2.resize(img, (img_scale_h, img_scale_w))
-        img = mmcv.imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
+        img = my_imnormalize(img, mean=np.array([123.675, 116.28, 103.53]), std=np.array([58.395, 57.12, 57.375]), to_rgb=True)
         preprocessed_img = np.expand_dims(img.transpose(2, 0, 1), axis=0)
         return preprocessed_img, ori_shape
 

app/service/prompt_generation/chatgpt_for_translation.py

@@ -90,7 +90,7 @@ def get_response(messages):
 
 def get_translation_from_llama3(text):
     start_time = time.time()
-    url = "http://10.1.1.240:11434/api/generate"
+    url = f"http://{settings.A6000_SERVICE_HOST}:12434/api/generate"
     # url = "http://10.1.1.240:1143/api/generate"
 
     # prompt = f"System: {prefix_for_llama}\nUser:[{text}]"
@@ -103,8 +103,8 @@ def get_translation_from_llama3(text):
 
     # 创建请求的负载 translator是自定义的翻译模型
     payload = {
-        "model": "translator",
+        "model": "AiDA-translator:latest",
-        "prompt": f"[{text}]",
+        "prompt": text,
         "stream": False
     }
     # 将负载转换为 JSON 格式
@@ -148,7 +148,7 @@ def get_translation_from_llama3(text):
 def get_prompt_from_image(image_path, text):
     start_time = time.time()
     # url = "http://localhost:11434/api/generate"
-    url = "http://10.1.1.243:11434/api/generate"
+    url = f"http://{settings.B_4_X_4090_SERVICE_HOST}:11434/api/generate"
 
     image_base64 = minio_util.minio_url_to_base64(image_path.img)
     # image_base64 = minio_url_to_base64(image_path)
@@ -180,7 +180,7 @@ def get_prompt_from_image(image_path, text):
 
 def main():
     """Main function"""
-    text = get_translation_from_llama3("[火焰]")
+    text = get_translation_from_llama3("火焰")
     print(text)
 
 

app/service/recommendation_system/config.py

@@ -14,7 +14,7 @@ REDIS_KEY_USER_PREF_PREFIX = "user_pref"
 RECOMMENDATION_CONFIG = {
     # 时间衰减半衰期（用于计算时间衰减权重）
     # 值越小，最近的行为权重越大
-    "K_half": 20,
+    "K_half": 10,
 
     # 探索与利用的比例 (0.0-1.0)
     # - 值越大，使用探索分支（随机推荐）的几率越大，结果更随机
@@ -25,7 +25,7 @@ RECOMMENDATION_CONFIG = {
     # 向量检索返回的候选数量
     # 值越大，候选池越大，但计算成本也越高
     # 建议范围: 100-1000
-    "topk": 1000,
+    "topk": 200,
 
     # Style 加分系数（同 style 的候选进行加分）
     # 值越大，匹配 style 的候选被选中的概率越大
@@ -53,7 +53,7 @@ RECOMMENDATION_CONFIG = {
 }
 
 # 数据库表名
-TABLE_USER_PREFERENCE_LOG = "user_preference_log_test"
+TABLE_USER_PREFERENCE_LOG = "user_preference"
 TABLE_SYS_FILE = "t_sys_file"
 
 # MySQL 连接配置（用于推荐系统）

app/service/recommendation_system/incremental_listener.py

@@ -1,6 +1,6 @@
 """
 增量监听模块
-实时监听 user_preference_log_test 表的新增记录，更新用户偏好向量
+实时监听 user_preference 表的新增记录，更新用户偏好向量
 """
 import logging
 import math
@@ -48,7 +48,7 @@ class IncrementalListener:
             if self.last_process_time is None:
                 # 第一次运行，查询最近30分钟的数据
                 cursor.execute(f"""
-                    SELECT id, account_id, path, category, style, data_time, is_system_sketch, sys_file_id
+                    SELECT id, account_id, path, category, style, data_time
                     FROM {TABLE_USER_PREFERENCE_LOG}
                     WHERE data_time > DATE_SUB(NOW(), INTERVAL 30 MINUTE)
                     ORDER BY data_time
@@ -56,7 +56,7 @@ class IncrementalListener:
             else:
                 # 基于上次处理时间查询
                 cursor.execute(f"""
-                    SELECT id, account_id, path, category, style, data_time, is_system_sketch, sys_file_id
+                    SELECT id, account_id, path, category, style, data_time
                     FROM {TABLE_USER_PREFERENCE_LOG}
                     WHERE data_time > %s
                     ORDER BY data_time
@@ -258,7 +258,7 @@ class IncrementalListener:
                     }
                 else:
                     # 用户图
-                    # 从 user_preference_log_test 获取 category（如果有）
+                    # 从 user_preference 获取 category（如果有）
                     cursor.execute(f"""
                         SELECT category 
                         FROM {TABLE_USER_PREFERENCE_LOG} 
@@ -308,6 +308,10 @@ class IncrementalListener:
 
 def start_background_listener(scheduler: BackgroundScheduler):
     """将增量监听任务注册到后台调度器"""
+    # 降低 apscheduler 的日志级别，避免大量刷屏
+    logging.getLogger('apscheduler.executors.default').setLevel(logging.WARNING)
+    logging.getLogger('apscheduler.scheduler').setLevel(logging.WARNING)
+    
     listener = IncrementalListener()
     scheduler.add_job(
         listener.process_once,

app/service/recommendation_system/milvus_client.py

@@ -23,7 +23,7 @@ def get_milvus_client() -> MilvusClient:
             _milvus_client = MilvusClient(
                 uri=settings.MILVUS_URL,
                 token=settings.MILVUS_TOKEN,
-                db_name=settings.MILVUS_DB,
+                db_name="",
             )
             logger.info("Milvus 客户端连接成功")
         except Exception as e:
@@ -203,39 +203,74 @@ def search_similar_vectors(
         query_vector: np.ndarray,
         category: str,
         topk: int = 500,
-        style: Optional[str] = None
+        style: Optional[str] = None,
+        style_boost_ratio: float = 0.2
 ) -> List[Dict]:
     """
     向量相似度检索
-    
+
     Args:
         query_vector: 查询向量（2048维）
         category: 类别过滤
         topk: 返回数量
-        style: 风格过滤（可选）
+        style: 风格过滤（可选）- 当提供时，会给对应style的结果加分
-    
+        style_boost_ratio: 风格加分比例（默认0.1，即10%）
+
     Returns:
         检索结果列表，每个元素包含 path, score, style, category 等字段
     """
     client = get_milvus_client()
 
     try:
-        # 构建过滤表达式
+        # 如果没有指定style，使用原始逻辑
-        # 使用 filter 参数而不是 expr（根据 pymilvus MilvusClient API）
+        if not style:
-        filter_expr = f"category == '{category}' && deprecated == 0"
+            filter_expr = f"category == '{category}' && deprecated == 0"
-        if style:
+            results = client.search(
-            filter_expr += f" && style == '{style}'"
+                collection_name=MILVUS_COLLECTION_SKETCH_VECTORS,
+                data=[query_vector.tolist()],
+                anns_field="feature_vector",
+                search_params={"metric_type": "IP", "params": {"nprobe": 10}},
+                limit=topk,
+                filter=filter_expr,
+                output_fields=["path", "style", "category", "sys_file_id"]
+            )
+        else:
+            # 有style参数时，使用两阶段搜索策略
 
-        # 搜索
+            # 第一阶段：搜索匹配style的向量，使用boosted query vector
-        results = client.search(
+            filter_expr_style = f"category == '{category}' && deprecated == 0 && style == '{style}'"
-            collection_name=MILVUS_COLLECTION_SKETCH_VECTORS,
+            boosted_query = query_vector * (1 + style_boost_ratio)
-            data=[query_vector.tolist()],
+            results_style = client.search(
-            anns_field="feature_vector",
+                collection_name=MILVUS_COLLECTION_SKETCH_VECTORS,
-            search_params={"metric_type": "IP", "params": {"nprobe": 10}},
+                data=[boosted_query.tolist()],
-            limit=topk,
+                anns_field="feature_vector",
-            filter=filter_expr,
+                search_params={"metric_type": "IP", "params": {"nprobe": 10}},
-            output_fields=["path", "style", "category", "sys_file_id"]
+                limit=topk,
-        )
+                filter=filter_expr_style,
+                output_fields=["path", "style", "category", "sys_file_id"]
+            )
+
+            # 第二阶段：搜索其他style的向量
+            filter_expr_others = f"category == '{category}' && deprecated == 0 && style != '{style}'"
+            results_others = client.search(
+                collection_name=MILVUS_COLLECTION_SKETCH_VECTORS,
+                data=[query_vector.tolist()],
+                anns_field="feature_vector",
+                search_params={"metric_type": "IP", "params": {"nprobe": 10}},
+                limit=topk,
+                filter=filter_expr_others,
+                output_fields=["path", "style", "category", "sys_file_id"]
+            )
+
+            # 合并结果
+            results = []
+            if results_style and len(results_style) > 0:
+                results.extend(results_style[0])
+            if results_others and len(results_others) > 0:
+                results.extend(results_others[0])
+
+            # 转换为单个结果列表格式
+            results = [results] if results else []
 
         # 格式化结果
         formatted_results = []
@@ -249,7 +284,10 @@ def search_similar_vectors(
                     "sys_file_id": hit.get("entity", {}).get("sys_file_id")
                 })
 
-        return formatted_results
+        # 按分数排序并返回topk
+        formatted_results.sort(key=lambda x: x["score"], reverse=True)
+        return formatted_results[:topk]
+
     except Exception as e:
         logger.error(f"向量检索失败: {e}", exc_info=True)
         return []
@@ -280,7 +318,7 @@ def query_random_candidates(category: str, style: Optional[str] = None, limit: i
             collection_name=MILVUS_COLLECTION_SKETCH_VECTORS,
             filter=filter_expr,
             output_fields=["path", "style", "category"],
-            limit=10000  # 先查询大量数据，然后随机选择
+            limit=10000
         )
 
         # 随机选择

app/service/recommendation_system/precompute.py

@@ -6,6 +6,7 @@ import logging
 import math
 import pymysql
 import numpy as np
+from datetime import datetime
 from typing import List, Dict, Tuple, Optional
 from collections import defaultdict
 
@@ -25,7 +26,7 @@ logger = logging.getLogger(__name__)
 
 def optimize_database_table():
     """
-    优化 user_preference_log_test 表结构
+    优化 user_preference 表结构
     添加冗余字段和索引
     """
     conn = None
@@ -317,8 +318,8 @@ def precompute_system_sketch_vectors(batch_size: int = 1000, retry_times: int =
 def compute_user_preference_vector(
     account_id: int,
     category: str,
-    conn: Optional[pymysql.connections.Connection] = None
+    conn: Optional[pymysql.connections.Connection] = None,
-    # max_date: Optional[datetime] = None
+    max_date: Optional[datetime] = None
 ) -> Optional[np.ndarray]:
     """
     计算用户偏好向量
@@ -419,8 +420,8 @@ def compute_user_preference_vector(
             p_i = 1 + math.log(1 + like_count)
             
             # 综合权重
-            # w_i = d_k * p_i
+            w_i = d_k * p_i
-            w_i = p_i
+            # w_i = p_i
 
             vectors.append(feature_vector)
             weights.append(w_i)
@@ -518,16 +519,16 @@ def run_precompute():
     logger.info("=" * 50)
     
     # 1. 优化数据库表结构
-    logger.info("\n[1/5] 优化数据库表结构...")
+    # logger.info("\n[1/5] 优化数据库表结构...")
-    optimize_database_table()
+    # optimize_database_table()
     
     # # 2. 创建 Milvus 集合
     # logger.info("\n[2/5] 创建 Milvus 集合...")
     # create_collection()
     
     # 3. 历史数据迁移
-    logger.info("\n[3/5] 历史数据迁移...")
+    # logger.info("\n[3/5] 历史数据迁移...")
-    migrate_historical_data()
+    # migrate_historical_data()
     
     # # 4. 系统图向量预计算
     # logger.info("\n[4/5] 系统图向量预计算...")
@@ -543,13 +544,13 @@ def run_precompute():
 
 
 if __name__ == "__main__":
-    # 1. 优化数据库表结构
+    # # 1. 优化数据库表结构
-    logger.info("\n[1/5] 优化数据库表结构...")
+    # logger.info("\n[1/5] 优化数据库表结构...")
-    optimize_database_table()
+    # optimize_database_table()
-
+    #
-    # 3. 历史数据迁移
+    # # 3. 历史数据迁移
-    logger.info("\n[3/5] 历史数据迁移...")
+    # logger.info("\n[3/5] 历史数据迁移...")
-    migrate_historical_data()
+    # migrate_historical_data()
 
     # 5. 初始用户偏好向量生成
     logger.info("\n[5/5] 初始用户偏好向量生成...")

app/service/recommendation_system/vector_utils.py

@@ -10,7 +10,7 @@ from torchvision import models, transforms
 from PIL import Image
 from minio import Minio
 
-from app.core.config import MINIO_URL, MINIO_ACCESS, MINIO_SECRET, MINIO_SECURE
+from app.core.config import settings
 from app.service.recommendation_system.config import RECOMMENDATION_CONFIG
 
 logger = logging.getLogger(__name__)
@@ -48,10 +48,10 @@ def get_minio_client():
     global _minio_client
     if _minio_client is None:
         _minio_client = Minio(
-            MINIO_URL,
+            settings.MINIO_URL,
-            access_key=MINIO_ACCESS,
+            access_key=settings.MINIO_ACCESS,
-            secret_key=MINIO_SECRET,
+            secret_key=settings.MINIO_SECRET,
-            secure=MINIO_SECURE
+            secure=settings.MINIO_SECURE
         )
     return _minio_client
 

app/service/utils/image_normalize.py

@@ -0,0 +1,27 @@
+import cv2
+import numpy as np
+
+
+def my_imnormalize(img, mean, std, to_rgb=True):
+    """Inplace normalize an image with mean and std.
+
+    Args:
+        img (ndarray): Image to be normalized.
+        mean (ndarray): The mean to be used for normalize.
+        std (ndarray): The std to be used for normalize.
+        to_rgb (bool): Whether to convert to rgb.
+
+    Returns:
+        ndarray: The normalized image.
+    """
+    # cv2 inplace normalization does not accept uint8
+    img = img.copy().astype(np.float32)
+
+    assert img.dtype != np.uint8
+    mean = np.float64(mean.reshape(1, -1))
+    stdinv = 1 / np.float64(std.reshape(1, -1))
+    if to_rgb:
+        cv2.cvtColor(img, cv2.COLOR_BGR2RGB, img)  # inplace
+    cv2.subtract(img, mean, img)  # inplace
+    cv2.multiply(img, stdinv, img)  # inplace
+    return img

app/service/utils/new_oss_client.py

@@ -81,7 +81,7 @@ if __name__ == '__main__':
     # url = "aida-users/89/sketchboard/female/Dress/e6724ab7-8d3f-4677-abe0-c3e42ab7af85.jpeg"
     # url = "aida-users/87/print/956614a2-7e75-4fbe-9ed0-c1831e37a2c9-4-87.png"
     # url = "aida-users/89/single_logo/123-89.png"
-    url = "lanecarford/lc_stylist_agent_outfit_items/141/ee25ec85-d504-4b42-9a18-db6682fe9e3b-6.jpg"
+    url = "aida-results/result_a7adcbd8-ef8d-11f0-8c92-0966ede33ab5.png"
 
     # url = "aida-collection-element/12148/Sketchboard/95ea577b-305b-4a62-b30a-39c0dd3ddb3f.png"
     read_type = "2"

app/service/utils/redis_utils.py

@@ -91,6 +91,21 @@ class Redis(object):
         r = cls._get_r()
         r.expire(name, expire_in_seconds)
 
+    @classmethod
+    def scan_keys(cls, pattern="*"):
+        """
+        扫描匹配模式的key
+        """
+        r = cls._get_r()
+        keys = []
+        cursor = 0
+        while True:
+            cursor, partial_keys = r.scan(cursor, match=pattern, count=1000)
+            keys.extend(partial_keys)
+            if cursor == 0:
+                break
+        return [key.decode('utf-8') if isinstance(key, bytes) else key for key in keys]
+
 
 if __name__ == '__main__':
     redis_client = Redis()

docker-compose.yml

@@ -1,13 +1,20 @@
 services:
   aida_server:
+    container_name: "AiDA_${SERVE_ENV}_Server"
     build:
       context: .
       dockerfile: Dockerfile
     working_dir: /app
     volumes:
       - ./app:/app/app
-      - ./.env_prod:/app/.env
+      - ./.env:/app/.env
       - /etc/localtime:/etc/localtime:ro
       - ./seg_cache:/seg_cache
     ports:
-      - "10200:80"
+      - "${SERVE_PORT}:80"
+    networks:
+      - aida_app_net
+networks:
+  aida_app_net:
+    external: true
+    name: aida_app_net

logging_env.py

@@ -1,10 +1,15 @@
+import os
+
 from app.core.config import settings
 
 LOGGER_CONFIG_DICT = {
     'version': 1,
     'disable_existing_loggers': False,
     'formatters': {
-        'simple': {'format': '%(asctime)s %(filename)s [line:%(lineno)d] %(levelname)s %(message)s'}
+        'simple': {
+            'format': '%(asctime)s %(filename)s [line:%(lineno)d] %(levelname)s %(message)s',
+            'datefmt': '%Y-%m-%d %H:%M:%S'  # 补充日期格式，日志更易读
+        }
     },
     'handlers': {
         'console': {
@@ -17,7 +22,7 @@ LOGGER_CONFIG_DICT = {
             'class': 'logging.handlers.RotatingFileHandler',
             'level': 'INFO',
             'formatter': 'simple',
-            'filename': f'{settings.LOGS_PATH}info.log',
+            'filename': os.path.join(settings.LOGS_PATH, 'info.log'),
             'maxBytes': 10485760,
             'backupCount': 50,
             'encoding': 'utf8',
@@ -26,7 +31,7 @@ LOGGER_CONFIG_DICT = {
             'class': 'logging.handlers.RotatingFileHandler',
             'level': 'ERROR',
             'formatter': 'simple',
-            'filename': f'{settings.LOGS_PATH}error.log',
+            'filename': os.path.join(settings.LOGS_PATH, 'error.log'),
             'maxBytes': 10485760,
             'backupCount': 20,
             'encoding': 'utf8',
@@ -35,7 +40,7 @@ LOGGER_CONFIG_DICT = {
             'class': 'logging.handlers.RotatingFileHandler',
             'level': 'DEBUG',
             'formatter': 'simple',
-            'filename': f'{settings.LOGS_PATH}debug.log',
+            'filename': os.path.join(settings.LOGS_PATH, 'debug.log'),
             'maxBytes': 10485760,
             'backupCount': 50,
             'encoding': 'utf8',
@@ -45,7 +50,7 @@ LOGGER_CONFIG_DICT = {
         'my_module': {'level': 'INFO', 'handlers': ['console'], 'propagate': 'no'}
     },
     'root': {
-        'level': 'INFO',
+        'level': 'DEBUG',
         'handlers': ['error_file_handler', 'info_file_handler', 'debug_file_handler', 'console'],
     },
 }

pyproject.toml

@@ -23,8 +23,8 @@ dependencies = [
     "load-dotenv>=0.1.0",
     "loguru>=0.7.3",
     "minio>=7.2.20",
-    "mmcv>=2.2.0",
     "moviepy==1.0.3",
+    "np>=1.0.2",
     "numpy<2",
     "ollama>=0.6.1",
     "opencv-python>=4.11.0.86",

uv.lock

@@ -8,15 +8,6 @@ resolution-markers = [
     "(platform_machine != 'aarch64' and sys_platform == 'linux') or (sys_platform != 'darwin' and sys_platform != 'linux')",
 ]
 
-[[package]]
-name = "addict"
-version = "2.4.0"
-source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/85/ef/fd7649da8af11d93979831e8f1f8097e85e82d5bfeabc8c68b39175d8e75/addict-2.4.0.tar.gz", hash = "sha256:b3b2210e0e067a281f5646c8c5db92e99b7231ea8b0eb5f74dbdf9e259d4e494", size = 9186, upload-time = "2020-11-21T16:21:31.416Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/6a/00/b08f23b7d7e1e14ce01419a467b583edbb93c6cdb8654e54a9cc579cd61f/addict-2.4.0-py3-none-any.whl", hash = "sha256:249bb56bbfd3cdc2a004ea0ff4c2b6ddc84d53bc2194761636eb314d5cfa5dfc", size = 3832, upload-time = "2020-11-21T16:21:29.588Z" },
-]
-
 [[package]]
 name = "agentaction"
 version = "0.1.7"
@@ -1671,43 +1662,6 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/3e/9a/b697530a882588a84db616580f2ba5d1d515c815e11c30d219145afeec87/minio-7.2.20-py3-none-any.whl", hash = "sha256:eb33dd2fb80e04c3726a76b13241c6be3c4c46f8d81e1d58e757786f6501897e", size = 93751, upload-time = "2025-11-27T00:37:13.993Z" },
 ]
 
-[[package]]
-name = "mmcv"
-version = "2.2.0"
-source = { registry = "https://pypi.org/simple" }
-dependencies = [
-    { name = "addict" },
-    { name = "mmengine" },
-    { name = "numpy" },
-    { name = "opencv-python" },
-    { name = "packaging" },
-    { name = "pillow" },
-    { name = "pyyaml" },
-    { name = "regex", marker = "sys_platform == 'win32'" },
-    { name = "yapf" },
-]
-sdist = { url = "https://files.pythonhosted.org/packages/e9/a2/57a733e7e84985a8a0e3101dfb8170fc9db92435c16afad253069ae3f9df/mmcv-2.2.0.tar.gz", hash = "sha256:ac479247e808d8802f89eadf04d4118de86bdfe81361ec5aed0cc1bf731c67c9", size = 479121, upload-time = "2024-04-24T14:24:28.064Z" }
-
-[[package]]
-name = "mmengine"
-version = "0.10.7"
-source = { registry = "https://pypi.org/simple" }
-dependencies = [
-    { name = "addict" },
-    { name = "matplotlib" },
-    { name = "numpy" },
-    { name = "opencv-python" },
-    { name = "pyyaml" },
-    { name = "regex", marker = "sys_platform == 'win32'" },
-    { name = "rich" },
-    { name = "termcolor" },
-    { name = "yapf" },
-]
-sdist = { url = "https://files.pythonhosted.org/packages/17/14/959360bbd8374e23fc1b720906999add16a3ac071a501636db12c5861ff5/mmengine-0.10.7.tar.gz", hash = "sha256:d20ffcc31127567e53dceff132612a87f0081de06cbb7ab2bdb7439125a69225", size = 378090, upload-time = "2025-03-04T12:23:09.568Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/98/8e/f98332248aad102511bea4ae19c0ddacd2f0a994f3ca4c82b7a369e0af8b/mmengine-0.10.7-py3-none-any.whl", hash = "sha256:262ac976a925562f78cd5fd14dd1bc9b680ed0aa81f0d85b723ef782f99c54ee", size = 452720, upload-time = "2025-03-04T12:23:06.339Z" },
-]
-
 [[package]]
 name = "mmh3"
 version = "5.2.0"
@@ -1801,6 +1755,12 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/9e/c9/b2622292ea83fbb4ec318f5b9ab867d0a28ab43c5717bb85b0a5f6b3b0a4/networkx-3.6.1-py3-none-any.whl", hash = "sha256:d47fbf302e7d9cbbb9e2555a0d267983d2aa476bac30e90dfbe5669bd57f3762", size = 2068504, upload-time = "2025-12-08T17:02:38.159Z" },
 ]
 
+[[package]]
+name = "np"
+version = "1.0.2"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/40/7d/749666e5a9976dcbc4d16d487bbe571efc6bbf4cdf3f4620c0ccc52b57ef/np-1.0.2.tar.gz", hash = "sha256:781265283f3823663ad8fb48741aae62abcf4c78bc19f908f8aa7c1d3eb132f8", size = 7419, upload-time = "2017-10-05T11:26:00.956Z" }
+
 [[package]]
 name = "numpy"
 version = "1.26.4"
@@ -2269,15 +2229,6 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/bc/96/aaa61ce33cc98421fb6088af2a03be4157b1e7e0e87087c888e2370a7f45/pillow-12.0.0-cp312-cp312-win_arm64.whl", hash = "sha256:7dfb439562f234f7d57b1ac6bc8fe7f838a4bd49c79230e0f6a1da93e82f1fad", size = 2436012, upload-time = "2025-10-15T18:22:23.621Z" },
 ]
 
-[[package]]
-name = "platformdirs"
-version = "4.5.1"
-source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/cf/86/0248f086a84f01b37aaec0fa567b397df1a119f73c16f6c7a9aac73ea309/platformdirs-4.5.1.tar.gz", hash = "sha256:61d5cdcc6065745cdd94f0f878977f8de9437be93de97c1c12f853c9c0cdcbda", size = 21715, upload-time = "2025-12-05T13:52:58.638Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/cb/28/3bfe2fa5a7b9c46fe7e13c97bda14c895fb10fa2ebf1d0abb90e0cea7ee1/platformdirs-4.5.1-py3-none-any.whl", hash = "sha256:d03afa3963c806a9bed9d5125c8f4cb2fdaf74a55ab60e5d59b3fde758104d31", size = 18731, upload-time = "2025-12-05T13:52:56.823Z" },
-]
-
 [[package]]
 name = "posthog"
 version = "5.4.0"
@@ -2746,17 +2697,6 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/2c/58/ca301544e1fa93ed4f80d724bf5b194f6e4b945841c5bfd555878eea9fcb/referencing-0.37.0-py3-none-any.whl", hash = "sha256:381329a9f99628c9069361716891d34ad94af76e461dcb0335825aecc7692231", size = 26766, upload-time = "2025-10-13T15:30:47.625Z" },
 ]
 
-[[package]]
-name = "regex"
-version = "2025.11.3"
-source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/cc/a9/546676f25e573a4cf00fe8e119b78a37b6a8fe2dc95cda877b30889c9c45/regex-2025.11.3.tar.gz", hash = "sha256:1fedc720f9bb2494ce31a58a1631f9c82df6a09b49c19517ea5cc280b4541e01", size = 414669, upload-time = "2025-11-03T21:34:22.089Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/59/9b/7c29be7903c318488983e7d97abcf8ebd3830e4c956c4c540005fcfb0462/regex-2025.11.3-cp312-cp312-win32.whl", hash = "sha256:3839967cf4dc4b985e1570fd8d91078f0c519f30491c60f9ac42a8db039be204", size = 266194, upload-time = "2025-11-03T21:31:51.53Z" },
-    { url = "https://files.pythonhosted.org/packages/1a/67/3b92df89f179d7c367be654ab5626ae311cb28f7d5c237b6bb976cd5fbbb/regex-2025.11.3-cp312-cp312-win_amd64.whl", hash = "sha256:e721d1b46e25c481dc5ded6f4b3f66c897c58d2e8cfdf77bbced84339108b0b9", size = 277069, upload-time = "2025-11-03T21:31:53.151Z" },
-    { url = "https://files.pythonhosted.org/packages/d7/55/85ba4c066fe5094d35b249c3ce8df0ba623cfd35afb22d6764f23a52a1c5/regex-2025.11.3-cp312-cp312-win_arm64.whl", hash = "sha256:64350685ff08b1d3a6fff33f45a9ca183dc1d58bbfe4981604e70ec9801bbc26", size = 270330, upload-time = "2025-11-03T21:31:54.514Z" },
-]
-
 [[package]]
 name = "requests"
 version = "2.32.5"
@@ -3224,8 +3164,8 @@ dependencies = [
     { name = "load-dotenv" },
     { name = "loguru" },
     { name = "minio" },
-    { name = "mmcv" },
     { name = "moviepy" },
+    { name = "np" },
     { name = "numpy" },
     { name = "ollama" },
     { name = "opencv-python" },
@@ -3275,8 +3215,8 @@ requires-dist = [
     { name = "load-dotenv", specifier = ">=0.1.0" },
     { name = "loguru", specifier = ">=0.7.3" },
     { name = "minio", specifier = ">=7.2.20" },
-    { name = "mmcv", specifier = ">=2.2.0" },
     { name = "moviepy", specifier = "==1.0.3" },
+    { name = "np", specifier = ">=1.0.2" },
     { name = "numpy", specifier = "<2" },
     { name = "ollama", specifier = ">=0.6.1" },
     { name = "opencv-python", specifier = ">=4.11.0.86" },
@@ -3605,18 +3545,6 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/54/85/6ec269b0952ec7e36ba019125982cf11d91256a778c7c3f98a4c5043d283/xxhash-3.6.0-cp312-cp312-win_arm64.whl", hash = "sha256:eae5c13f3bc455a3bbb68bdc513912dc7356de7e2280363ea235f71f54064829", size = 27876, upload-time = "2025-10-02T14:34:54.371Z" },
 ]
 
-[[package]]
-name = "yapf"
-version = "0.43.0"
-source = { registry = "https://pypi.org/simple" }
-dependencies = [
-    { name = "platformdirs" },
-]
-sdist = { url = "https://files.pythonhosted.org/packages/23/97/b6f296d1e9cc1ec25c7604178b48532fa5901f721bcf1b8d8148b13e5588/yapf-0.43.0.tar.gz", hash = "sha256:00d3aa24bfedff9420b2e0d5d9f5ab6d9d4268e72afbf59bb3fa542781d5218e", size = 254907, upload-time = "2024-11-14T00:11:41.584Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/37/81/6acd6601f61e31cfb8729d3da6d5df966f80f374b78eff83760714487338/yapf-0.43.0-py3-none-any.whl", hash = "sha256:224faffbc39c428cb095818cf6ef5511fdab6f7430a10783fdfb292ccf2852ca", size = 256158, upload-time = "2024-11-14T00:11:39.37Z" },
-]
-
 [[package]]
 name = "yarl"
 version = "1.22.0"