DAY38 Dataset和DataLoader

拉普拉斯妖1082025-12-17 18:51

python 复制代码

import torch 
import torch.nn as nn
import torch.optim as optim 
from torch.utils.data import DataLoader,Dataset
from torchvision import datasets,transforms
import matplotlib.pyplot as plt

torch.manual_seed(42)

python 复制代码

transform=transforms.Compose([
    transforms.ToTensor(),
    transforms.Normalize((0.1307,),(0.3081,))
])

python 复制代码

train_dataset=datasets.MNIST(
    root='./data',
    train=True,
    download=True,
    transform=transform
)

test_dataset=datasets.MNIST(
    root='./data',
    train=False,
    transform=transform
)

python 复制代码

import matplotlib.pyplot as plt
sample_idx=torch.randint(0,len(train_dataset),size=(1,)).item()
image,label=train_dataset[sample_idx]

python 复制代码

from torchvision import datasets, transforms
class MNIST(Dataset):
    def __init____init__(self,root,train=True,transform=None):
        self.data,self.targets=fetch_mnist_data(root,train)
        self.transform=transform

    def __len__(self):
        return len(self.data)
    
    def __getitem__(self,idx):
        img,target=self.data[idx],self.targets[idx]

        if self.transform is not None:
            img=self.transform(img)
    
        return img,target

python 复制代码

def imshow(img):
    img=img*0.3081+0.1307
    nping=img.numpy()
    plt.imshow(nping[0],cmap='gray')
    plt.show()

print(f"Label:{label}")
imshow(image)

python 复制代码

train_loader=DataLoader(
    train_dataset,
    batch_size=64,
    shuffle=True
)

test_loader=DataLoader(
    test_dataset,
    batch_size=1000
)

下载cifar数据集并获取其中一张图片

python 复制代码

import torchvision
import numpy as np
transform=transforms.Compose([
    transforms.ToTensor(),
    transforms.Normalize((0.5,0.5,0.5),(0.5,0.5,0.5))
])

train_dataset=torchvision.datasets.CIFAR10(
    root='./data',
    train=True,
    download=True,
    transform=transform
)

sample_idx=0
image,label=train_dataset[sample_idx]

# 类别名称
classes = ('plane', 'car', 'bird', 'cat', 'deer', 
           'dog', 'frog', 'horse', 'ship', 'truck')

print(f"图片形状: {image.shape}")
print(f"标签: {label} - {classes[label]}")

def imshow(img):
    img=img*0.5+0.5
    npimg=img.numpy()
    plt.imshow(np.transpose(npimg,(1,2,0)))
    plt.axis('off')

imshow(image)
plt.title(f'Label: {classes[label]} ({label})')
plt.show()