python数据增强——对数据预处理（随即裁剪、翻转）

import os
from PIL import Image
import torch.utils.data as data
import torchvision.transforms as transforms
import cv2
import numpy as np
import random

random.seed(2021)


class CamObjDataset(data.Dataset):
    def __init__(self, image_root, gt_root, edge_root, trainsize):
        self.trainsize = trainsize  # 用于训练时图像的大小
        # 图像 标签 边缘图像列表创建和排序
        self.images = [image_root + f for f in os.listdir(image_root) if f.endswith('.jpg')]
        self.gts = [gt_root + f for f in os.listdir(gt_root) if f.endswith('.png')]
        self.edges = [edge_root + f for f in os.listdir(edge_root) if f.endswith('.png')]
        # 对这些列表进行排序，以确保图像、标签和边缘图像的顺序一致
        self.images = sorted(self.images)
        self.gts = sorted(self.gts)
        self.edges = sorted(self.edges)
        # 过滤不匹配文件：移除尺寸不一致的图像、标签和边缘图像
        self.filter_files()
        self.size = len(self.images)
        self.kernel = np.ones((5, 5), np.uint8)
        # 用于图像和标签的转换操作，调整大小、转换为张量、标准化
        self.img_transform = transforms.Compose([
            transforms.Resize((self.trainsize, self.trainsize)),
            transforms.ToTensor(),
            transforms.Normalize([0.485, 0.456, 0.406],
                                 [0.229, 0.224, 0.225])])
        self.ge_transform = transforms.Compose([
            transforms.Resize((self.trainsize, self.trainsize)),
            transforms.ToTensor()])

    # 随机决定是否对图像进行水平翻转
    def getFlip(self):
        p = random.randint(0, 1)
        self.flip = transforms.RandomHorizontalFlip(p)

    # 按照索引获取数据集中的单个样本
    def __getitem__(self, index):
        self.getFlip()
        # 加载图像、标签和边缘图像
        image = self.rgb_loader(self.images[index])
        gt = self.binary_loader(self.gts[index])
        edge = cv2.imread(self.edges[index], cv2.IMREAD_GRAYSCALE)
        image = self.flip(image)
        image = self.img_transform(image)
        gt = self.flip(gt)
        gt = self.ge_transform(gt)
        edge = cv2.dilate(edge, self.kernel, iterations=1)
        edge = Image.fromarray(edge)
        edge = self.flip(edge)
        edge = self.ge_transform(edge)
        return image, gt, edge

    #  确保图像、标签和边缘图像尺寸一致
    def filter_files(self):
        assert len(self.images) == len(self.gts)
        images = []
        gts = []
        edges = []
        for img_path, gt_path, edge_path in zip(self.images, self.gts, self.edges):
            img = Image.open(img_path)
            gt = Image.open(gt_path)
            edge = Image.open(edge_path)
            if img.size == gt.size and img.size == edge.size:
                images.append(img_path)
                gts.append(gt_path)
                edges.append(edge_path)
        self.images = images
        self.gts = gts
        self.edges = edges

    # 用于加载RGB图像和二值图像
    def rgb_loader(self, path):
        with open(path, 'rb') as f:
            img = Image.open(f)
            return img.convert('RGB')

    def binary_loader(self, path):
        with open(path, 'rb') as f:
            img = Image.open(f)
            return img.convert('L')

    #  如果图像或标签小于trainsize，则调整其大小
    def resize(self, img, gt):
        assert img.size == gt.size
        w, h = img.size
        if h < self.trainsize or w < self.trainsize:
            h = max(h, self.trainsize)
            w = max(w, self.trainsize)
            return img.resize((w, h), Image.BILINEAR), gt.resize((w, h), Image.NEAREST)
        else:
            return img, gt

    def __len__(self):
        return self.size