深度学习数据集的准备

oldsweet老甜甜

于 2021-06-10 07:27:43 发布

阅读量202

点赞数

CC 4.0 BY-SA版权

文章标签：深度学习 tensorflow 二叉树

本文链接：https://blog.youkuaiyun.com/qq_45867029/article/details/117764798

本文介绍了一个使用PyTorch加载CIFAR10数据集并利用TensorBoard进行图像展示的示例。通过组合多个转换操作，如调整大小和将图像转换为张量，实现了数据预处理。同时，该示例还展示了如何通过DataLoader批量读取数据，并使用SummaryWriter将图像写入到TensorBoard中。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

import cv2
import torchvision.transforms as transforms
import torchvision.datasets as datasets
from torch.utils.data import DataLoader
from torch.utils.data import Dataset
from torch.utils.tensorboard import SummaryWriter

transform_to_tensor = transforms.ToTensor()
transform_size = transforms.Resize((32, 32))
transform_compose = transforms.Compose([transform_to_tensor, transform_size])
writer = SummaryWriter("logs")
if __name__ == '__main__':
    train_set = datasets.CIFAR10(root='./datasets', download=True, transform=transform_compose, train=True)
    test_set = datasets.CIFAR10(root='./datasets', download=True, transform=transform_compose, train=False)
    Data = DataLoader(train_set, batch_size=4, shuffle=True, num_workers=0, drop_last=True)
    step = 0
    for data in Data:
        imags, labels = data
        writer.add_images("images", imags, step)
        step = step + 1
    writer.close()