Aprendizaje profundo: carga de datos

0. Prefacio

marco: pytorch

1. Descargue el conjunto de datos directamente de Internet para la prueba del modelo.

    tsf = transforms.Compose([transforms.ToTensor()])
    train_data = datasets.FashionMNIST(root='./data/fashionmnist/train',train=True,transform=tsf,download=True)
    test_data = datasets.FashionMNIST(root='./data/fashionmnist/test',train=False,transform=tsf,download=True)
    train_iter = DataLoader(train_data,batch_size=32,shuffle=True)
    test_iter = DataLoader(test_data,batch_size=32,shuffle=False)

2. Cargar desde carpeta local

    tsf = transforms.Compose([transforms.Resize((28, 28)),
                              transforms.ToTensor()])
    train_data = datasets.ImageFolder(root=r'image\dogcat\dogcat\train',transform=tsf)
    train_iter = DataLoader(train_data,batch_size=32,shuffle=True)

Estructura de directorios de la carpeta local