diff --git a/app/service/design_batch/design_batch_celery.py b/app/service/design_batch/design_batch_celery.py index 06ccc5e..f5cdc58 100644 --- a/app/service/design_batch/design_batch_celery.py +++ b/app/service/design_batch/design_batch_celery.py @@ -5,9 +5,9 @@ from celery import Celery from minio import Minio from app.core.config import * -from app.service.design_batch.item import BodyItem, TopItem, BottomItem +from app.service.design_batch.item import BodyItem, TopItem, BottomItem, AccessoriesItem from app.service.design_batch.utils.MQ import publish_status -from app.service.design_batch.utils.organize import organize_body, organize_clothing +from app.service.design_batch.utils.organize import organize_body, organize_clothing, organize_accessories from app.service.design_batch.utils.save_json import oss_upload_json from app.service.design_batch.utils.synthesis_item import update_base_size_priority, synthesis, synthesis_single @@ -19,6 +19,8 @@ logging.getLogger('pika').setLevel(logging.WARNING) logger = logging.getLogger() minio_client = Minio(MINIO_URL, access_key=MINIO_ACCESS, secret_key=MINIO_SECRET, secure=MINIO_SECURE) +print("start") + def process_item(item, basic): # 处理project中单个item @@ -28,9 +30,14 @@ def process_item(item, basic): elif item['type'].lower() in ['blouse', 'outwear', 'dress', 'tops']: top_server = TopItem(data=item, basic=basic, minio_client=minio_client) item_data = top_server.process() - else: + elif item['type'].lower() in ['skirt', 'trousers', 'bottoms']: bottom_server = BottomItem(data=item, basic=basic, minio_client=minio_client) item_data = bottom_server.process() + elif item['type'].lower() in ['accessories']: + bottom_server = AccessoriesItem(data=item, basic=basic, minio_client=minio_client) + item_data = bottom_server.process() + else: + raise NotImplementedError(f"Item type {item['type']} not implemented") return item_data @@ -40,6 +47,10 @@ def process_layer(item, layers): body_layer = organize_body(item) layers.append(body_layer) return item['body_image'].size + elif item['name'] == 'accessories': + front_layer, back_layer = organize_accessories(item) + layers.append(front_layer) + layers.append(back_layer) else: front_layer, back_layer = organize_clothing(item) layers.append(front_layer) @@ -48,6 +59,9 @@ def process_layer(item, layers): @celery_app.task def batch_design(objects_data, tasks_id, json_name): + print(objects_data) + print(tasks_id) + print(json_name) object_response = [] threads = [] active_threads = 0 @@ -121,6 +135,7 @@ def batch_design(objects_data, tasks_id, json_name): for t in threads: t.join() logger.debug(object_response) + print(object_response) oss_upload_json(minio_client, object_response, json_name) publish_status(tasks_id, "ok", json_name) return object_response diff --git a/app/service/design_batch/item.py b/app/service/design_batch/item.py index cad1488..5ddfdc7 100644 --- a/app/service/design_batch/item.py +++ b/app/service/design_batch/item.py @@ -1,4 +1,4 @@ -from app.service.design_batch.pipeline import * +from app.service.design_fast.pipeline import LoadImage, KeyPoint, Segmentation, Color, PrintPainting, Scaling, Split, LoadBodyImage, ContourDetection class BaseItem: @@ -9,6 +9,27 @@ class BaseItem: self.result.update(basic) +class AccessoriesItem(BaseItem): + def __init__(self, data, basic, minio_client): + super().__init__(data, basic) + self.Accessories_pipeline = [ + LoadImage(minio_client), + # KeyPoint(), + ContourDetection(), + # Segmentation(minio_client), + # BackPerspective(minio_client), + Color(minio_client), + PrintPainting(minio_client), + Scaling(), + Split(minio_client) + ] + + def process(self): + for item in self.Accessories_pipeline: + self.result = item(self.result) + return self.result + + class TopItem(BaseItem): def __init__(self, data, basic, minio_client): super().__init__(data, basic) @@ -16,6 +37,7 @@ class TopItem(BaseItem): LoadImage(minio_client), KeyPoint(), Segmentation(minio_client), + # BackPerspective(minio_client), Color(minio_client), PrintPainting(minio_client), Scaling(), @@ -35,7 +57,8 @@ class BottomItem(BaseItem): LoadImage(minio_client), KeyPoint(), ContourDetection(), - # Segmentation(), + Segmentation(minio_client), + # BackPerspective(minio_client), Color(minio_client), PrintPainting(minio_client), Scaling(), diff --git a/app/service/design_batch/pipeline/__init__.py b/app/service/design_batch/pipeline/__init__.py index ec55933..f265bbe 100644 --- a/app/service/design_batch/pipeline/__init__.py +++ b/app/service/design_batch/pipeline/__init__.py @@ -1,3 +1,4 @@ +from .back_perspective import BackPerspective from .color import Color from .contour_detection import ContourDetection from .keypoint import KeyPoint @@ -13,6 +14,7 @@ __all__ = [ 'KeyPoint', 'ContourDetection', 'Segmentation', + 'BackPerspective', 'Color', 'PrintPainting', 'Scaling', diff --git a/app/service/design_batch/pipeline/back_perspective.py b/app/service/design_batch/pipeline/back_perspective.py new file mode 100644 index 0000000..5ddd37c --- /dev/null +++ b/app/service/design_batch/pipeline/back_perspective.py @@ -0,0 +1,79 @@ +import cv2 +import numpy as np + +from app.service.design_fast.utils.design_ensemble import get_seg_result +from app.service.utils.new_oss_client import oss_upload_image + + +class BackPerspective: + def __init__(self, minio_client): + self.minio_client = minio_client + + def __call__(self, result): + + # 如果sketch为系统图 查看是否有对应的 背后视角图 + if result['path'].split('/')[0] == 'aida-sys-image': + file_path = result['path'].replace("images", 'images_back', 1) + if self.is_file_exists(bucket_name='aida-sys-image', file_name=file_path[file_path.find('/') + 1:]): + result['back_perspective_url'] = file_path + return result + else: + seg_result = get_seg_result("1", result['image'])[0] + elif result['name'] in ['blouse', 'outwear', 'dress', 'tops']: + seg_result = result['seg_result'] + else: + seg_result = get_seg_result("1", result['image'])[0] + + m = self.thicken_contours_and_display(seg_result, thickness=10, color=(0, 0, 0)) + back_sketch = result['image'].copy() + back_sketch[m > 100] = 255 + # 上传背后视角图 + _, img_encoded = cv2.imencode(".jpg", back_sketch) + + resp = oss_upload_image(self.minio_client, bucket='test', object_name=result['path'], image_bytes=img_encoded.tobytes()) + result['back_perspective_url'] = f"{resp.bucket_name}/{resp.object_name}" + return result + + def thicken_contours_and_display(self, mask, thickness=10, color=(0, 0, 0)): + mask = mask.astype(np.uint8) * 255 + # 查找轮廓 + contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE) + + # 创建一个彩色副本用于绘制轮廓 + mask_color = cv2.cvtColor(mask, cv2.COLOR_GRAY2BGR) + + def thicken_contour_inward(contour, thick): + # 创建一个空白的黑色图像与原始掩码大小相同 + blank = np.zeros_like(mask) + # 在空白图像上绘制白色的轮廓 + cv2.drawContours(blank, [contour], -1, 255, thickness=thick) + # 找到轮廓的中心（可以用重心等方法近似） + M = cv2.moments(contour) + cx = int(M['m10'] / M['m00']) + cy = int(M['m01'] / M['m00']) + # 进行距离变换，离中心越近的值越小 + dist_transform = cv2.distanceTransform(255 - blank, cv2.DIST_L2, 5) + # 根据距离变换的值来决定是否保留像素，离中心近的像素更容易被保留 + result = np.zeros_like(mask) + for i in range(dist_transform.shape[0]): + for j in range(dist_transform.shape[1]): + if dist_transform[i, j] < thick: + result[i, j] = 255 + return result + + for contour in contours: + thickened_contour = thicken_contour_inward(contour, thickness) + mask_color[thickened_contour > 0] = color + + _, binary_result = cv2.threshold(mask_color, 127, 255, cv2.THRESH_BINARY) + + # 转换为掩码形式 + mask_result = cv2.cvtColor(binary_result, cv2.COLOR_BGR2GRAY) + return mask_result + + def is_file_exists(self, bucket_name, file_name): + try: + self.minio_client.stat_object(bucket_name, file_name) + return True + except Exception: + return False diff --git a/app/service/design_batch/pipeline/color.py b/app/service/design_batch/pipeline/color.py index 546c671..d6c84e4 100644 --- a/app/service/design_batch/pipeline/color.py +++ b/app/service/design_batch/pipeline/color.py @@ -14,14 +14,39 @@ class Color: def __call__(self, result): dim_image_h, dim_image_w = result['image'].shape[0:2] + # 渐变色 if "gradient" in result.keys() and result['gradient'] != "": bucket_name = result['gradient'].split('/')[0] object_name = result['gradient'][result['gradient'].find('/') + 1:] pattern = self.get_gradient(bucket_name=bucket_name, object_name=object_name) resize_pattern = cv2.resize(pattern, (dim_image_w, dim_image_h), interpolation=cv2.INTER_AREA) + # 无色 + elif "color" not in result.keys() or result['color'] == "": + result['final_image'] = result['pattern_image'] = result['single_image'] = result['image'] + result['alpha'] = 100 / 255.0 + return result + # 正常颜色 else: pattern = self.get_pattern(result['color']) resize_pattern = cv2.resize(pattern, (dim_image_w, dim_image_h), interpolation=cv2.INTER_AREA) + + if "partial_color" in result.keys() and result['partial_color'] != "": + bucket_name = result['partial_color'].split('/')[0] + object_name = result['partial_color'][result['partial_color'].find('/') + 1:] + partial_color = oss_get_image(oss_client=self.minio_client, bucket=bucket_name, object_name=object_name, data_type="cv2") + h, w = partial_color.shape[0:2] + resize_pattern = cv2.resize(resize_pattern, (w, h), interpolation=cv2.INTER_AREA) + # 分离出 png 图的 alpha 通道 + alpha_channel = partial_color[:, :, 3] + # 提取 png 图的 RGB 通道 + png_rgb = partial_color[:, :, :3] + # 创建一个与 cv 图大小相同的掩码，用于指示哪些像素需要替换 + mask = alpha_channel > 0 + # 将掩码扩展为 3 通道，以便与 cv 图进行逐元素操作 + mask_3ch = np.stack([mask] * 3, axis=-1) + # 根据掩码将 png 图的颜色覆盖到 cv 图上 + resize_pattern[mask_3ch] = png_rgb[mask_3ch] + resize_pattern = cv2.resize(resize_pattern, (dim_image_w, dim_image_h), interpolation=cv2.INTER_AREA) closed_mo = np.expand_dims(result['mask'], axis=2).repeat(3, axis=2) gray_mo = np.expand_dims(result['gray'], axis=2).repeat(3, axis=2) get_image_fir = resize_pattern * (closed_mo / 255) * (gray_mo / 255) diff --git a/app/service/design_batch/pipeline/keypoint.py b/app/service/design_batch/pipeline/keypoint.py index 313a613..73d7586 100644 --- a/app/service/design_batch/pipeline/keypoint.py +++ b/app/service/design_batch/pipeline/keypoint.py @@ -4,7 +4,8 @@ import numpy as np from pymilvus import MilvusClient from app.core.config import * -from app.service.design_batch.utils.design_ensemble import get_keypoint_result +from app.service.design_fast.utils.design_ensemble import get_keypoint_result +from app.service.utils.decorator import ClassCallRunTime, RunTime logger = logging.getLogger(__name__) @@ -16,14 +17,15 @@ class KeyPoint: def get_name(cls): return cls.name + @ClassCallRunTime def __call__(self, result): if result['name'] in ['blouse', 'skirt', 'dress', 'outwear', 'trousers', 'tops', 'bottoms']: # 查询是否有数据 且类别相同 相同则直接读 不同则推理后更新 # result['clothes_keypoint'] = self.infer_keypoint_result(result) site = 'up' if result['name'] in ['blouse', 'outwear', 'dress', 'tops'] else 'down' # keypoint_cache = search_keypoint_cache(result["image_id"], site) - keypoint_cache = self.keypoint_cache(result, site) + # keypoint_cache = self.keypoint_cache(result, site) + keypoint_cache = False # 取消向量查询 直接过模型推理 - # keypoint_cache = False if keypoint_cache is False: keypoint_infer_result, site = self.infer_keypoint_result(result) result['clothes_keypoint'] = self.save_keypoint_cache(result["image_id"], keypoint_infer_result, site) @@ -87,7 +89,7 @@ class KeyPoint: 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): try: client = MilvusClient(uri=MILVUS_URL, token=MILVUS_TOKEN, db_name=MILVUS_ALIAS) diff --git a/app/service/design_batch/pipeline/loading.py b/app/service/design_batch/pipeline/loading.py index 8f02378..5a55d9d 100644 --- a/app/service/design_batch/pipeline/loading.py +++ b/app/service/design_batch/pipeline/loading.py @@ -1,6 +1,9 @@ +import io import logging import cv2 +import numpy as np +from PIL import Image from app.service.utils.new_oss_client import oss_get_image @@ -71,6 +74,8 @@ class LoadImage: keypoint = 'head_point' elif name == 'earring': keypoint = 'ear_point' + elif name == 'accessories': + keypoint = "accessories" else: raise KeyError(f"{name} does not belong to item category list: blouse, outwear, dress, trousers, skirt, " f"bag, shoes, hairstyle, earring.") diff --git a/app/service/design_batch/pipeline/print_painting.py b/app/service/design_batch/pipeline/print_painting.py index 6fe40d8..1534f9c 100644 --- a/app/service/design_batch/pipeline/print_painting.py +++ b/app/service/design_batch/pipeline/print_painting.py @@ -15,8 +15,25 @@ class PrintPainting: single_print = result['print']['single'] overall_print = result['print']['overall'] element_print = result['print']['element'] + partial_path = result['print']['partial'] if 'partial' in result['print'] else None result['single_image'] = None result['print_image'] = None + # TODO 给result['pattern_image'] resize 到resize_scale的大小 + # TODO 给result['mask'] resize 到resize_scale的大小 + + if result['resize_scale'][0] == 1.0 and result['resize_scale'][1] == 1.0: + pass + else: + height, width = result['pattern_image'].shape[:2] + new_width = int(width * result['resize_scale'][0]) + new_height = int(height * result['resize_scale'][1]) + + result['pattern_image'] = cv2.resize(result['pattern_image'], (new_width, new_height)) + result['final_image'] = cv2.resize(result['final_image'], (new_width, new_height)) + result['mask'] = cv2.resize(result['mask'], (new_width, new_height)) + result['gray'] = cv2.resize(result['gray'], (new_width, new_height)) + + print(1) if overall_print['print_path_list']: painting_dict = {'dim_image_h': result['pattern_image'].shape[0], 'dim_image_w': result['pattern_image'].shape[1]} result['print_image'] = result['pattern_image'] @@ -39,7 +56,7 @@ class PrintPainting: for i in range(len(single_print['print_path_list'])): image, image_mode = self.read_image(single_print['print_path_list'][i]) if image_mode == "RGBA": - new_size = (int(image.width * single_print['print_scale_list'][i]), int(image.height * single_print['print_scale_list'][i])) + new_size = (int(result['pattern_image'].shape[1] * single_print['print_scale_list'][i][0]), int(result['pattern_image'].shape[0] * single_print['print_scale_list'][i][1])) mask = image.split()[3] resized_source = image.resize(new_size) @@ -62,9 +79,12 @@ class PrintPainting: 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], single_print['print_scale_list'][i]) - rotate_image, rotated_new_size = self.img_rotate(image, single_print['print_angle_list'][i], single_print['print_scale_list'][i]) + 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]) @@ -143,7 +163,7 @@ class PrintPainting: for i in range(len(element_print['element_path_list'])): image, image_mode = self.read_image(element_print['element_path_list'][i]) if image_mode == "RGBA": - new_size = (int(image.width * element_print['element_scale_list'][i]), int(image.height * element_print['element_scale_list'][i])) + new_size = (int(result['final_image'].shape[1] * element_print['element_scale_list'][i][0]), int(result['final_image'].shape[0] * element_print['element_scale_list'][i][1])) mask = image.split()[3] resized_source = image.resize(new_size) @@ -165,9 +185,11 @@ class PrintPainting: 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, element_print['element_angle_list'][i], element_print['element_scale_list'][i]) - rotate_image, rotated_new_size = self.img_rotate(image, element_print['element_angle_list'][i], element_print['element_scale_list'][i]) + rotate_mask, _ = self.img_rotate(mask, element_print['element_angle_list'][i]) + rotate_image, rotated_new_size = self.img_rotate(image, element_print['element_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(element_print['location'][i][0] - rotated_new_size[0]), int(element_print['location'][i][1] - rotated_new_size[1]) @@ -241,6 +263,45 @@ class PrintPainting: temp_fg = np.expand_dims(result['mask'], axis=2).repeat(3, axis=2) tmp2 = (result['final_image'] * (temp_fg / 255)).astype(np.uint8) result['single_image'] = cv2.add(tmp1, tmp2) + + if partial_path: + 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) + image, image_mode = self.read_image(partial_path) + if image_mode == "RGBA": + new_size = (result['pattern_image'].shape[1], result['pattern_image'].shape[0]) + + mask = image.split()[3] + resized_source = image.resize(new_size) + resized_source_mask = mask.resize(new_size) + + # rotated_resized_source = resized_source.rotate(-partial_print['print_angle_list'][i]) + # rotated_resized_source_mask = resized_source_mask.rotate(-partial_print['print_angle_list'][i]) + + source_image_pil = Image.fromarray(cv2.cvtColor(print_background, cv2.COLOR_BGR2RGB)) + source_image_pil_mask = Image.fromarray(cv2.cvtColor(mask_background, cv2.COLOR_BGR2RGB)) + + source_image_pil.paste(resized_source, (0, 0), resized_source) + source_image_pil_mask.paste(resized_source_mask, (0, 0), resized_source_mask) + + 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) + 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) + tmp1 = (canvas * (temp_bg / 255)).astype(np.uint8) + temp_fg = np.expand_dims(result['mask'], axis=2).repeat(3, axis=2) + tmp2 = (result['final_image'] * (temp_fg / 255)).astype(np.uint8) + result['single_image'] = cv2.add(tmp1, tmp2) return result @staticmethod @@ -360,10 +421,10 @@ class PrintPainting: return print_image def get_print(self, print_dict): - if 'print_scale_list' not in print_dict.keys() or print_dict['print_scale_list'][0] < 0.3: + if 'print_scale_list' not in print_dict.keys() or print_dict['print_scale_list'][0][0] < 0.3: print_dict['scale'] = 0.3 else: - print_dict['scale'] = print_dict['print_scale_list'][0] + print_dict['scale'] = print_dict['print_scale_list'][0][0] bucket_name = print_dict['print_path_list'][0].split("/", 1)[0] object_name = print_dict['print_path_list'][0].split("/", 1)[1] @@ -386,8 +447,9 @@ class PrintPainting: # 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 - x_offset = print_w - int(location[0][1] % print_w) - y_offset = print_w - int(location[0][0] % print_h) + # 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]) @@ -409,7 +471,7 @@ class PrintPainting: return high, low @staticmethod - def img_rotate(image, angel, scale): + def img_rotate(image, angel): """顺时针旋转图像任意角度 Args: @@ -424,7 +486,7 @@ class PrintPainting: center = (w // 2, h // 2) # if type(angel) is not int: # angel = 0 - M = cv2.getRotationMatrix2D(center, -angel, scale) + M = cv2.getRotationMatrix2D(center, -angel, 1) # 调整旋转后的图像长宽 rotated_h = int((w * np.abs(M[0, 1]) + (h * np.abs(M[0, 0])))) rotated_w = int((h * np.abs(M[0, 1]) + (w * np.abs(M[0, 0])))) @@ -433,7 +495,7 @@ class PrintPainting: # 旋转图像 rotated_img = cv2.warpAffine(image, M, (rotated_w, rotated_h)) - return rotated_img, ((rotated_img.shape[1] - image.shape[1] * scale) // 2, (rotated_img.shape[0] - image.shape[0] * scale) // 2) + return rotated_img, ((rotated_img.shape[1] - image.shape[1]) // 2, (rotated_img.shape[0] - image.shape[0]) // 2) # return rotated_img, (0, 0) @staticmethod @@ -442,8 +504,11 @@ class PrintPainting: angle: 旋转的角度 crop: 是否需要进行裁剪，布尔向量 """ + if not isinstance(crop, bool): + raise ValueError("The 'crop' parameter must be a boolean.") + crop_image = lambda img, x0, y0, w, h: img[y0:y0 + h, x0:x0 + w] - w, h = img.shape[:2] + h, w = img.shape[:2] # 旋转角度的周期是360° angle %= 360 # 计算仿射变换矩阵 @@ -455,7 +520,7 @@ class PrintPainting: if crop: # 裁剪角度的等效周期是180° angle_crop = angle % 180 - if angle > 90: + if angle_crop > 90: angle_crop = 180 - angle_crop # 转化角度为弧度 theta = angle_crop * np.pi / 180 diff --git a/app/service/design_batch/pipeline/scale.py b/app/service/design_batch/pipeline/scale.py index 1908a9c..d1c7a36 100644 --- a/app/service/design_batch/pipeline/scale.py +++ b/app/service/design_batch/pipeline/scale.py @@ -46,4 +46,16 @@ class Scaling: result['scale'] = result['scale_bag'] elif result['keypoint'] == 'ear_point': result['scale'] = result['scale_earrings'] + elif result['keypoint'] == 'accessories': + # 由于没有识别配饰keypoint的模型 所以统一将配饰的两个关键点设定为 (0,0) (0,img.width) + # 模特的关键点设定为(0,0) (0,320/2) 距离比例简写为 160 / img.width + distance_clo = result['img_shape'][1] + distance_bdy = 320 / 2 + + if distance_clo == 0: + result['scale'] = 1 + else: + result['scale'] = distance_bdy / distance_clo + else: + result['scale'] = 1 return result diff --git a/app/service/design_batch/pipeline/segmentation.py b/app/service/design_batch/pipeline/segmentation.py index aa05c0d..0c9c51e 100644 --- a/app/service/design_batch/pipeline/segmentation.py +++ b/app/service/design_batch/pipeline/segmentation.py @@ -5,7 +5,8 @@ import cv2 import numpy as np from app.core.config import SEG_CACHE_PATH -from app.service.design_batch.utils.design_ensemble import get_seg_result +from app.service.design_fast.utils.design_ensemble import get_seg_result +from app.service.utils.decorator import ClassCallRunTime from app.service.utils.new_oss_client import oss_get_image logger = logging.getLogger() @@ -15,6 +16,7 @@ class Segmentation: def __init__(self, minio_client): self.minio_client = minio_client + @ClassCallRunTime def __call__(self, result): if "seg_mask_url" in result.keys() and result['seg_mask_url'] != "": seg_mask = oss_get_image(oss_client=self.minio_client, bucket=result['seg_mask_url'].split('/')[0], object_name=result['seg_mask_url'][result['seg_mask_url'].find('/') + 1:], data_type="cv2") @@ -31,24 +33,37 @@ class Segmentation: result['back_mask'] = np.array(green_mask, dtype=np.uint8) * 255 result['mask'] = result['front_mask'] + result['back_mask'] else: - # 本地查询seg 缓存是否存在 - _, seg_result = self.load_seg_result(result["image_id"]) - result['seg_result'] = seg_result - if not _: + # preview 过模型 不缓存 + if "preview_submit" in result.keys() and result['preview_submit'] == "preview": # 推理获得seg 结果 - seg_result = get_seg_result(result["image_id"], result['image'])[0] + seg_result = get_seg_result(result["image_id"], result['image']) + # submit 过模型 缓存 + elif "preview_submit" in result.keys() and result['preview_submit'] == "submit": + # 推理获得seg 结果 + seg_result = get_seg_result(result["image_id"], result['image']) 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 + 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 + if not _ or result["image"].shape[:2] != seg_result.shape: + # 推理获得seg 结果 + seg_result = get_seg_result(result["image_id"], result['image']) + self.save_seg_result(seg_result, result['image_id']) + result['seg_result'] = seg_result + # 处理前片后片 - temp_front = seg_result == 1.0 + temp_front = seg_result == 1 result['front_mask'] = (255 * (temp_front + 0).astype(np.uint8)) - temp_back = seg_result == 2.0 + temp_back = seg_result == 2 result['back_mask'] = (255 * (temp_back + 0).astype(np.uint8)) result['mask'] = result['front_mask'] + result['back_mask'] return result @staticmethod def save_seg_result(seg_result, image_id): - file_path = f"seg_cache/{image_id}.npy" + file_path = f"{SEG_CACHE_PATH}{image_id}.npy" try: np.save(file_path, seg_result) logger.debug(f"保存成功 ：{os.path.abspath(file_path)}") @@ -57,7 +72,7 @@ class Segmentation: @staticmethod def load_seg_result(image_id): - file_path = f"seg_cache/{image_id}.npy" + file_path = f"{SEG_CACHE_PATH}{image_id}.npy" # logger.info(f"load seg file name is :{SEG_CACHE_PATH}{image_id}.npy") try: seg_result = np.load(file_path) diff --git a/app/service/design_batch/pipeline/split.py b/app/service/design_batch/pipeline/split.py index 5dbcef5..88e8e75 100644 --- a/app/service/design_batch/pipeline/split.py +++ b/app/service/design_batch/pipeline/split.py @@ -7,10 +7,11 @@ from PIL import Image from cv2 import cvtColor, COLOR_BGR2RGBA from app.core.config import AIDA_CLOTHING -from app.service.design_batch.utils.conversion_image import rgb_to_rgba -from app.service.design_batch.utils.upload_image import upload_png_mask +from app.service.design_fast.utils.conversion_image import rgb_to_rgba +from app.service.design_fast.utils.transparent import sketch_to_transparent +from app.service.design_fast.utils.upload_image import upload_png_mask from app.service.utils.generate_uuid import generate_uuid -from app.service.utils.new_oss_client import oss_upload_image +from app.service.utils.new_oss_client import oss_upload_image, oss_get_image class Split(object): @@ -20,51 +21,95 @@ class Split(object): def __call__(self, result): try: - if result['name'] in ('outwear', 'dress', 'blouse', 'skirt', 'trousers', 'tops', 'bottoms'): - front_mask = result['front_mask'] - back_mask = result['back_mask'] + if result['name'] in ('outwear', 'dress', 'blouse', 'skirt', 'trousers', 'tops', 'bottoms', 'accessories'): + + if result['resize_scale'][0] == 1.0 and result['resize_scale'][1] == 1.0: + front_mask = result['front_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)) + back_mask = cv2.resize(result['back_mask'], (new_width, new_height)) + rgba_image = rgb_to_rgba(result['final_image'], front_mask + back_mask) - new_size = (int(rgba_image.shape[1] * result["scale"] * result["resize_scale"][0]), int(rgba_image.shape[0] * result["scale"] * result["resize_scale"][1])) + new_size = (int(rgba_image.shape[1] * result["scale"]), int(rgba_image.shape[0] * result["scale"])) rgba_image = cv2.resize(rgba_image, new_size) result_front_image = np.zeros_like(rgba_image) front_mask = cv2.resize(front_mask, new_size) result_front_image[front_mask != 0] = rgba_image[front_mask != 0] result_front_image_pil = Image.fromarray(cvtColor(result_front_image, 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_NEAREST) + # 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 = front_mask.shape mask_image = np.zeros((height, width, 3)) mask_image[front_mask != 0] = [0, 0, 255] - if result["name"] in ('blouse', 'dress', 'outwear', 'tops'): - result_back_image = np.zeros_like(rgba_image) - back_mask = cv2.resize(back_mask, new_size) - result_back_image[back_mask != 0] = rgba_image[back_mask != 0] - result_back_image_pil = Image.fromarray(cvtColor(result_back_image, 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] + # if result["name"] in ('blouse', 'dress', 'outwear', 'tops'): + # result_back_image = np.zeros_like(rgba_image) + # back_mask = cv2.resize(back_mask, new_size) + # result_back_image[back_mask != 0] = rgba_image[back_mask != 0] + # result_back_image_pil = Image.fromarray(cvtColor(result_back_image, 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] + # + # rbga_mask = rgb_to_rgba(mask_image, front_mask + back_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 + # 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 - rbga_mask = rgb_to_rgba(mask_image, front_mask + back_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 - 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) + result_back_image[back_mask != 0] = rgba_image[back_mask != 0] + result_back_image_pil = Image.fromarray(cvtColor(result_back_image, 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] + + rbga_mask = rgb_to_rgba(mask_image, front_mask + back_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 # 创建中间图层 result_pattern_image_rgba = rgb_to_rgba(result['pattern_image'], result['mask']) result_pattern_image_pil = Image.fromarray(cvtColor(result_pattern_image_rgba, COLOR_BGR2RGBA)) diff --git a/app/service/design_batch/utils/organize.py b/app/service/design_batch/utils/organize.py index 8190de0..33edc4f 100644 --- a/app/service/design_batch/utils/organize.py +++ b/app/service/design_batch/utils/organize.py @@ -33,8 +33,8 @@ def organize_clothing(layer): mask=cv2.resize(layer['mask'], layer["front_image"].size), gradient_string=layer['gradient_string'] if 'gradient_string' in layer.keys() else "", pattern_image_url=layer['pattern_image_url'], - pattern_image=layer['pattern_image'] - + pattern_image=layer['pattern_image'], + # back_perspective_url=layer['back_perspective_url'] if 'back_perspective_url' in layer.keys() else "" ) # 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), @@ -50,6 +50,46 @@ def organize_clothing(layer): mask=cv2.resize(layer['mask'], layer["front_image"].size), gradient_string=layer['gradient_string'] if 'gradient_string' in layer.keys() else "", pattern_image_url=layer['pattern_image_url'], + # back_perspective_url=layer['back_perspective_url'] if 'back_perspective_url' in layer.keys() else "" + ) + return front_layer, back_layer + + +def organize_accessories(layer): + # 起始坐标 + start_point = (0, 0) + # 前片数据 + 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"], + # mask_image=layer['front_mask_image'], + image_url=layer['front_image_url'], + mask_url=layer['mask_url'], + 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_image_url=layer['pattern_image_url'], + pattern_image=layer['pattern_image'], + # back_perspective_url=layer['back_perspective_url'] if 'back_perspective_url' in layer.keys() else "" + ) + # 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), + name=f'{layer["name"].lower()}_back', + image=layer["back_image"], + # mask_image=layer['back_mask_image'], + image_url=layer['back_image_url'], + mask_url=layer['mask_url'], + 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_image_url=layer['pattern_image_url'], + # back_perspective_url=layer['back_perspective_url'] if 'back_perspective_url' in layer.keys() else "" ) return front_layer, back_layer