img 应该是 PIL Image.得到了 <class 'torch.Te

img should be PIL Image. Got lt;class #39;torch.Tensor#39;gt;(img 应该是 PIL Image.得到了 lt;class torch.Tensorgt;)

问题描述

我正在尝试遍历加载器以检查它是否正常工作，但是给出了以下错误:

I'm trying to iterate through a loader to check if it's working, however the below error is given:

TypeError: img 应该是 PIL Image.得到了

我已经尝试添加 transforms.ToTensor() 和 transforms.ToPILImage() 并且它给了我一个错误要求相反.即，使用 ToPILImage()，它将要求张量，反之亦然.

I've tried adding both transforms.ToTensor() and transforms.ToPILImage() and it gives me an error asking for the opposite. i.e, with ToPILImage(), it will ask for tensor, and vice versa.

# Imports here
%matplotlib inline
import matplotlib.pyplot as plt
from torch import nn, optim
import torch.nn.functional as F
import torch
from torchvision import transforms, datasets, models
import seaborn as sns
import pandas as pd
import numpy as np

data_dir = 'flowers'
train_dir = data_dir + '/train'
valid_dir = data_dir + '/valid'
test_dir = data_dir + '/test'

#Creating transform for training set
train_transforms = transforms.Compose(
[transforms.Resize(255), 
transforms.CenterCrop(224), 
transforms.ToTensor(), 
transforms.RandomHorizontalFlip(), 
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])

#Creating transform for test set
test_transforms = transforms.Compose(
[transforms.Resize(255),
transforms.CenterCrop(224), 
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406],[0.229, 0.224, 0.225])])

#transforming for all data
train_data = datasets.ImageFolder(train_dir, transform=train_transforms)
test_data = datasets.ImageFolder(test_dir, transform = test_transforms)
valid_data = datasets.ImageFolder(valid_dir, transform = test_transforms)

#Creating data loaders for test and training sets
trainloader = torch.utils.data.DataLoader(train_data, batch_size = 32, 
shuffle = True)
testloader = torch.utils.data.DataLoader(test_data, batch_size=32)
images, labels = next(iter(trainloader))

它应该允许我在运行 plt.imshow(images[0]) 后简单地看到图像，如果它工作正常.

It should allow me to simply see the image once I run plt.imshow(images[0]), if its working correctly.

推荐答案

transforms.RandomHorizo​​ntalFlip() 适用于 PIL.Images，而不是 torch.Tensor.在上面的代码中，您在 transforms.RandomHorizo​​ntalFlip() 之前应用 transforms.ToTensor()，这会产生张量.

train_transforms = transforms.Compose([transforms.Resize(255), 
                                       transforms.CenterCrop(224),  
                                       transforms.RandomHorizontalFlip(),
                                       transforms.ToTensor(), 
                                       transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])