AiDA_Python/app/service/image2sketch_2/server.py

import cv2
import numpy
import numpy as np
import torch
import torch.nn as nn
import torchvision.transforms as transforms
from PIL import Image

from app.service.utils.oss_client import oss_get_image, oss_upload_image

norm_layer = nn.InstanceNorm2d

weights = [(0.7, 0.3), (0.5, 0.5), (0.3, 0.7), (0.1, 0.9), (0, 1)]
kernel = np.ones((3, 3), np.uint8)


class ResidualBlock(nn.Module):
    def __init__(self, in_features):
        super(ResidualBlock, self).__init__()

        conv_block = [nn.ReflectionPad2d(1),
                      nn.Conv2d(in_features, in_features, 3),
                      norm_layer(in_features),
                      nn.ReLU(inplace=True),
                      nn.ReflectionPad2d(1),
                      nn.Conv2d(in_features, in_features, 3),
                      norm_layer(in_features)
                      ]

        self.conv_block = nn.Sequential(*conv_block)

    def forward(self, x):
        return x + self.conv_block(x)


class Generator(nn.Module):
    def __init__(self, input_nc, output_nc, n_residual_blocks=9, sigmoid=True):
        super(Generator, self).__init__()

        # Initial convolution block
        model0 = [nn.ReflectionPad2d(3),
                  nn.Conv2d(input_nc, 64, 7),
                  norm_layer(64),
                  nn.ReLU(inplace=True)]
        self.model0 = nn.Sequential(*model0)

        # Downsampling
        model1 = []
        in_features = 64
        out_features = in_features * 2
        for _ in range(2):
            model1 += [nn.Conv2d(in_features, out_features, 3, stride=2, padding=1),
                       norm_layer(out_features),
                       nn.ReLU(inplace=True)]
            in_features = out_features
            out_features = in_features * 2
        self.model1 = nn.Sequential(*model1)

        model2 = []
        # Residual blocks
        for _ in range(n_residual_blocks):
            model2 += [ResidualBlock(in_features)]
        self.model2 = nn.Sequential(*model2)

        # Upsampling
        model3 = []
        out_features = in_features // 2
        for _ in range(2):
            model3 += [nn.ConvTranspose2d(in_features, out_features, 3, stride=2, padding=1, output_padding=1),
                       norm_layer(out_features),
                       nn.ReLU(inplace=True)]
            in_features = out_features
            out_features = in_features // 2
        self.model3 = nn.Sequential(*model3)

        # Output layer
        model4 = [nn.ReflectionPad2d(3),
                  nn.Conv2d(64, output_nc, 7)]
        if sigmoid:
            model4 += [nn.Sigmoid()]

        self.model4 = nn.Sequential(*model4)

    def forward(self, x, cond=None):
        out = self.model0(x)
        out = self.model1(out)
        out = self.model2(out)
        out = self.model3(out)
        out = self.model4(out)

        return out


model1 = Generator(3, 1, 3)
model1.load_state_dict(torch.load('app/service/image2sketch_2/model.pth', map_location=torch.device('cpu')))
model1.eval()


def predict(input_img, width):
    transform = transforms.Compose([transforms.Resize(width, Image.BICUBIC), transforms.ToTensor()])
    input_img = transform(input_img)
    input_img = torch.unsqueeze(input_img, 0)

    with torch.no_grad():
        drawing = model1(input_img)[0].detach()

    drawing = transforms.ToPILImage()(drawing)

    # 转ndarray
    drawing = numpy.array(drawing)
    return drawing


def get_image(image_url):
    image = oss_get_image(bucket=image_url.split('/')[0], object_name=image_url[image_url.find('/') + 1:], data_type="PIL")
    image = image.convert('RGB')
    width = image.size[0]
    height = image.size[1]
    return image, width, height


def processing_pipeline(image_url, thickness, sketch_bucket, sketch_name):
    thickness = int(thickness)
    # 提取sketch
    image, width, height = get_image(image_url)
    sketch_image = predict(image, width)

    # 设定线条粗细
    if thickness != 0:
        dilated = cv2.erode(sketch_image, kernel, iterations=1)
        # 将原图与膨胀后的图像进行混合，使用不同的权重
        sketch_image = cv2.addWeighted(sketch_image, weights[thickness][0], dilated, weights[thickness][1], 0)

    #    上传minio
    image_bytes = cv2.imencode(".jpg", sketch_image)[1].tobytes()
    req = oss_upload_image(bucket=sketch_bucket, object_name=sketch_name, image_bytes=image_bytes)
    return f"{req.bucket_name}/{req.object_name}"


if __name__ == '__main__':
    result_url = processing_pipeline("aida-users/89/relight_image/d5f0d967-f8e8-424d-98f9-a8ad8313deec-0-89.png", 1, "test", "test123.jpg")
    print(result_url)