快捷導(dǎo)航

使用python/pytorch讀取數(shù)據(jù)集的示例代碼

更新時間：2023年12月29日 11:38:07 作者：jedi-knight

這篇文章主要為大家詳細介紹了使用python/pytorch讀取數(shù)據(jù)集的示例,文中的示例代碼講解詳細,具有一定參考價值,感興趣的小伙伴可以跟隨小編一起學(xué)習(xí)一下

MNIST數(shù)據(jù)集

MNIST數(shù)據(jù)集包含了6萬張手寫數(shù)字（[1,28,28]尺寸），以特殊格式存儲。本文首先將MNIST數(shù)據(jù)集另存為png格式，然后再讀取png格式圖片，開展后續(xù)訓(xùn)練

另存為png格式

import torch
from torch.utils.data import Dataset
from torchvision.datasets import MNIST
from torch.utils.data import DataLoader
from tqdm import tqdm
from torchvision import models, transforms
from torchvision.utils import save_image
from PIL import Image

#將MNIST數(shù)據(jù)集轉(zhuǎn)換為圖片
tf = transforms.Compose([transforms.ToTensor()]) # mnist is already normalised 0 to 1
datasetMNIST = MNIST("./data", train=True, download=True, transform=tf)
pbar = tqdm(datasetMNIST)
for index, (img,cl) in enumerate(pbar):
   save_image(img, f"./data/MNIST_PNG/x/{index}.png")
   # 以寫入模式打開文件
   with open(f"./data/MNIST_PNG/c/{index}.txt", "w", encoding="utf-8") as file:
        # 將字符串寫入文件
        file.write(f"{cl}")

注意：MNIST源數(shù)據(jù)存放在./data文件下，如果沒有數(shù)據(jù)也沒關(guān)系，代碼會自動從網(wǎng)上下載。另存為png的數(shù)據(jù)放在了./data/MNIST_PNG/文件下。子文件夾x存放6萬張圖片，子文件夾c存放6萬個文本文件，每個文本文件內(nèi)有一行字符串，說明該對應(yīng)的手寫數(shù)字是幾（標簽）。

讀取png格式數(shù)據(jù)集

class MyMNISTDataset(Dataset):
   def __init__(self, data):
       self.data = data

   def __len__(self):
       return len(self.data)

   def __getitem__(self, idx):
       x = self.data[idx][0] #圖像
       y = self.data[idx][1] #標簽
       return x, y
   
def load_data(dataNum=60000):
    data = []
    pbar = tqdm(range(dataNum))
    for i in pbar:
        # 指定圖片路徑
        image_path = f'./data/MNIST_PNG/x/{i}.png'
        cond_path=f'./data/MNIST_PNG/c/{i}.txt'
        # 定義圖像預(yù)處理
        preprocess = transforms.Compose([
        transforms.Grayscale(num_output_channels=1),  # 將圖像轉(zhuǎn)換為灰度圖像（單通道）
        transforms.ToTensor()
        ])
        # 使用預(yù)處理加載圖像
        image_tensor = preprocess(Image.open(image_path))
        # 加載條件文檔（tag）
        with open(cond_path, 'r') as file:
            line = file.readline()
            number = int(line)  # 將字符串轉(zhuǎn)換為整數(shù)，圖像的類別
            data.append((image_tensor, number))
    return data
   

data=load_data(60000)
# 創(chuàng)建數(shù)據(jù)集實例
dataset = MyMNISTDataset(data)

# 創(chuàng)建數(shù)據(jù)加載器
dataloader = DataLoader(dataset, batch_size=4, shuffle=True)
pbar = tqdm(dataloader)

for index, (img,cond) in enumerate(pbar):
    #這里對每一批進行訓(xùn)練...
    print(f"Batch {index}: img = {img.shape}, cond = {cond}")

load_data函數(shù)用于讀取數(shù)據(jù)文件，返回一個data張量。data張量又被用于構(gòu)造MyMNISTDataset類的對象dataset，dataset對象又被DataLoader函數(shù)轉(zhuǎn)換為dataloader。

dataloader事實上按照batch將數(shù)據(jù)集進行了分割，4張圖片一組進行訓(xùn)練。上述代碼的輸出如下：

......
Batch 7847: img = torch.Size([4, 1, 28, 28]), cond = tensor([0, 1, 5, 2])
Batch 7848: img = torch.Size([4, 1, 28, 28]), cond = tensor([2, 2, 6, 0])
Batch 7849: img = torch.Size([4, 1, 28, 28]), cond = tensor([4, 3, 0, 9])
Batch 7850: img = torch.Size([4, 1, 28, 28]), cond = tensor([6, 2, 9, 5])
Batch 7851: img = torch.Size([4, 1, 28, 28]), cond = tensor([7, 2, 4, 4])
Batch 7852: img = torch.Size([4, 1, 28, 28]), cond = tensor([1, 4, 2, 6])
Batch 7853: img = torch.Size([4, 1, 28, 28]), cond = tensor([2, 5, 3, 5])
Batch 7854: img = torch.Size([4, 1, 28, 28]), cond = tensor([7, 1, 0, 1])
Batch 7855: img = torch.Size([4, 1, 28, 28]), cond = tensor([9, 8, 9, 7])
Batch 7856: img = torch.Size([4, 1, 28, 28]), cond = tensor([4, 6, 6, 7])
Batch 7857: img = torch.Size([4, 1, 28, 28]), cond = tensor([7, 4, 1, 6])
Batch 7858: img = torch.Size([4, 1, 28, 28]), cond = tensor([5, 4, 6, 5])
Batch 7859: img = torch.Size([4, 1, 28, 28]), cond = tensor([6, 3, 1, 9])
Batch 7860: img = torch.Size([4, 1, 28, 28]), cond = tensor([5, 5, 8, 6])
Batch 7861: img = torch.Size([4, 1, 28, 28]), cond = tensor([0, 4, 8, 9])
Batch 7862: img = torch.Size([4, 1, 28, 28]), cond = tensor([2, 3, 5, 8])
Batch 7863: img = torch.Size([4, 1, 28, 28]), cond = tensor([8, 0, 0, 6])
......

到此這篇關(guān)于使用python/pytorch讀取數(shù)據(jù)集的示例代碼的文章就介紹到這了,更多相關(guān)python/pytorch讀取數(shù)據(jù)集內(nèi)容請搜索腳本之家以前的文章或繼續(xù)瀏覽下面的相關(guān)文章希望大家以后多多支持腳本之家！

您可能感興趣的文章: