前言
从官网上下载的是长这个样子的
想看图片，咋办咧，看下面代码
import torch
import torchvision
import numpy as np
import os
import cv2
batch_size = 50transform_predict = torchvision.transforms.Compose([torchvision.transforms.ToTensor(),
])
#-----#
# train 为True 则是解压出训练图片 为Fasle的时候则解压出测试图片
#------#
image_data = torchvision.datasets.CIFAR10(root='/home/netted/img_process_ml/temp', train=True, download=False, transform=transform_predict)
image_loader = torch.utils.data.DataLoader(image_data, batch_size, shuffle=True, num_workers=0)path = '/home/netted/img_process_ml/temp/train'
os.makedirs(path,exist_ok=True)
for i in range(10):os.makedirs(f'{path}/{i}',exist_ok=True)def format(image):image = image.clone().detach().cpu().squeeze(0)image = np.around(image.mul(255))image = np.uint8(image).transpose(1, 2, 0)return imagedef data(image_loader):idx0 = 0idx1 = 0idx2 = 0idx3 = 0idx4 = 0idx5 = 0idx6 = 0idx7 = 0idx8 = 0idx9 = 0for i, (data, target) in enumerate(image_loader):for idx in range(len(data)):label = target[idx].item()image = format(data[idx])if label == 0:cv2.imwrite(f'{path}/{label}/plane_{idx0}.png',image)idx0 += 1if label == 1:cv2.imwrite(f'{path}/{label}/car_{idx1}.png', image)idx1 += 1if label == 2:cv2.imwrite(f'{path}/{label}/bird_{idx2}.png', image)idx2 += 1if label == 3:cv2.imwrite(f'{path}/{label}/cat_{idx3}.png', image)idx3 += 1if label == 4:cv2.imwrite(f'{path}/{label}/deer_{idx4}.png', image)idx4 += 1if label == 5:cv2.imwrite(f'{path}/{label}/dog_{idx5}.png', image)idx5 += 1if label == 6:cv2.imwrite(f'{path}/{label}/frog_{idx6}.png', image)idx6 += 1if label == 7:cv2.imwrite(f'{path}/{label}/horse_{idx7}.png', image)idx7 += 1if label == 8:cv2.imwrite(f'{path}/{label}/ship_{idx8}.png', image)idx8 += 1if label == 9:cv2.imwrite(f'{path}/{label}/truck_{idx9}.png', image)idx9 += 1data(image_loader)然后就解压出来了
当然可以自行调整将它们都合在一个文件夹里面，个人喜好
原包与自己生成好的链接如下：
链接：https://pan.baidu.com/s/1pkAFVjZ2f3ibPvMe4TtjOQ?pwd=noia
提取码：noia
欢迎大家点赞或收藏~
可以鼓励作者加快更新哟~
