TypeError: 'builtin_function_or_method' object is not iterable 报错解决

爱心觉罗

在用Pytorch做图像分类的时候，遇到了这个BUG，因为这段代码和网上例子一样仍报错，所以很奇怪，求解答
先贴上我报错的信息和代码

1. TypeError: 'builtin_function_or_method' object is not iterable

复制代码

1. import torch
   
2. import torch.utils.data as Data
   
3. import torchvision
   
4. from torchvision import transforms, datasets
   
5. 
   
6. # hyper parameters
   
7. BATCH_SIZE = 8
   
8. 
   
9. path = "DogsVSCats"
   
10. # define the data transform
    
11. data_transform = transforms.Compose([
    
12.     transforms.ToTensor(),  # change the pixel to [0, 1.0]
    
13.     transforms.CenterCrop(224),
    
14.     transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
    
15. ])
    
16. 
    
17. # define the data set in dictionary form
    
18. image_data = {x:datasets.ImageFolder(root=os.path.join(path, x), transform=data_transform)
    
19.               for x in ['train', 'valid']}
    
20. # define the data loader
    
21. data_loader = {x:Data.DataLoader(dataset=image_data[x], batch_size=BATCH_SIZE, shuffle=True)
    
22.                for x in ['train', 'valid']}
    
23. 
    
24. # preview a batch of data in data set
    
25. # x_example, y_example = next(iter(data_loader['train']))
    
26. x_train, y_train = next(iter(data_loader["train"]))
    
27. print("x_example个数:", len(x_train))
    
28. print("y_example个数:", len(y_train))

复制代码

报错的是第三行

1. x_train, y_train = next(iter(data_loader["train"]))

复制代码

求大神告知

AI的世界

定义transform时，一定要调整图片的大小，再ToTensor()
修改后的部分如下：调整CenterCrop()和ToTensor()的顺序

1. # define the data transform !!!参数顺序很重要！！！先调整大小再ToTensor
   
2. data_transform = transforms.Compose([
   
3.     transforms.CenterCrop(224),
   
4.     transforms.ToTensor(),  # change the pixel to [0, 1.0]
   
5.     transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
   
6. ])
   

复制代码