深度学习之图像分割数据集：卫星道路图像分割数据集（划分了训练集和测试集）

import cv2 import numpy as np import os import random from PIL import Image import matplotlib.pyplot as plt # https://blog.csdn.net/qq_44886601/article/details/135764589 # 将mask绘制在原图 def draw_image(im, ms, brg, opacity): image_mask = im.copy() contours, _ = cv2.findContours(ms, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE) # 查找轮廓 image_mask = cv2.drawContours(image_mask, contours, -1, (0, 255, 0), 2) # 绘制边界 image_mask = cv2.fillPoly(image_mask, contours, color=brg) # 填充 img_bgr = cv2.addWeighted(im, opacity, image_mask, 1 - opacity, 0) return im, ms, img_bgr def main(imagePath, labelPath, bgr, opacity): image = np.array(Image.open(imagePath).convert('RGB')) label = np.array(Image.open(labelPath).convert('L')) a, b, c = draw_image(image, label, bgr, opacity) plt.figure(figsize=(12, 8)) for index, i in enumerate((a, b, c)): plt.subplot(1, 3, index + 1) plt.imshow(i) plt.savefig('./result.png') # plt.show() if __name__ == '__main__': root = './data/train/images' images_path = [os.path.join(root, i) for i in os.listdir(root)] r = random.randint(0, len(images_path) - 1) img_path = images_path[r] # 随机取出一张图片 mask_path = img_path.replace('images', 'masks') # opacity 越小，掩膜效果越深 main(imagePath=img_path, labelPath=mask_path, bgr=(0, 0, 255), opacity=0.5)