Pytorch framework learning record 4 - the use of datasets (torchvision.dataset)

Pytorch框架学习记录4——数据集的使用（torchvision.dataset）

1. 数据集

在pytorch官网中我们可以看到pytorchThe situation of the data set it is equipped with,and the type of the dataset、使用方法等.在这里,We choose the smaller datasetCIFAR10as our example dataset.

The calling and usage code of this dataset is as follows：

torchvision.datasets.CIFAR10(root: str, train: bool = True, transform: Optional[Callable] = None, target_transform: Optional[Callable] = None, download: bool = False)

参数说明：

• root ( string ) – 数据集的根目录, cifar-10-batches-py如果下载设置为 True,则该目录存在或将保存到该目录.
• train ( bool,optional ) – 如果为真,则从训练集创建数据集,否则从测试集创建.
• transform ( callable,optional ) – 一个函数/转换,它接受 PIL 图像并返回转换后的版本.例如,transforms.RandomCrop
• target_transform ( callable,optional ) – 接收目标并对其进行转换的函数/转换.
• download ( bool,optional ) – 如果为 true,则从 Internet 下载数据集并将其放在根目录中.如果数据集已经下载,则不会再次下载.

2. 使用实例

下载CIFAR10数据集后,Convert its type to tensor类型,并在tensorboard中进行展示.

import torchvision
from torch.utils.tensorboard import SummaryWriter
from torchvision.transforms import transforms

dataset_transform = transforms.Compose([
    transforms.ToTensor()
])
train_set = torchvision.datasets.CIFAR10(root='./dataset', train=True, transform=dataset_transform, download=True)
test_set = torchvision.datasets.CIFAR10(root='./dataset', train=False, transform=dataset_transform, download=True)

writer = SummaryWriter('logs')

for i in range(10):
    img, label = train_set[i]
    writer.add_image('train10', img, i)

writer.close()

此外,You can also download directly from the link using your browser,下载完毕后,Also name one in the current directorydatasetfolder and put in,The above code does not make any changes,Manually downloaded datasets are automatically decompressed and corrected.