yjy20041202: FileNotFoundError: [Errno 2] No such file or directory: 'D:\\yiliao project\\UNet_Demo\\VOCdevkit\\VOC2007/JPEGImages\\142-2_jpg.rf.2dd296358cf41d5ad11b774aeafac2e4_mask.jpg'换成绝对路径了，还是报错了

视觉研坊: 在测试的后处理部分添加下面代码，代码主体不变，你参考着修改：[code=python] import numpy as np import cv2 from collections import Counter # 假设你已经有了预测结果，这是一个示例结果 # 这里的预测结果应该是你的U-Net模型的输出 # 比如：prediction = model.predict(test_image) # 这里使用随机生成的示例数据 height, width = 256, 256 num_classes = 5 # 假设有5个类别 prediction = np.random.randint(0, num_classes, (height, width)) # 方法一：使用Counter统计类别个数 def count_classes_with_counter(prediction): flat_prediction = prediction.flatten() class_counts = Counter(flat_prediction) return class_counts # 方法二：使用NumPy统计类别个数 def count_classes_with_numpy(prediction, num_classes): class_counts = np.zeros(num_classes, dtype=int) for i in range(num_classes): class_counts[i] = np.sum(prediction == i) return class_counts # 统计类别个数 class_counts_counter = count_classes_with_counter(prediction) class_counts_numpy = count_classes_with_numpy(prediction, num_classes) # 打印结果 print("Using Counter:") for class_label, count in class_counts_counter.items(): print(f"Class [/code]

视觉研坊: 没看懂你想问的问题，你都已经转为我博文格式了，后续按照博文中方法操作即可

视觉研坊: 你给的路径格式不对，重新导入

视觉研坊: 路径要么绝对路径：F：XXX\\VOCdevkit\\VOC2007/JPEGImages\\4-1_jpg.rf.acd9e78702c02f38062323e674dcf006_mask.jpg 要么相对路径VOCdevkit/VOC2007/JPEGImages/4-1_jpg.rf.acd9e78702c02f38062323e674dcf006_mask.jpg