《动手学深度学习 Pytorch版》 3.5 图像分类数据集

%matplotlib inline
import torch
import torchvision
from torch.utils import data
from torchvision import transforms
from d2l import torch as d2l

d2l.use_svg_display()

3.5.1 读取数据集

# 使用比 MNIST 数据集类似但更复杂的 Fashion-MINIST 数据集

trans = transforms.ToTensor()  # 将PILImage格式或者numpy.array格式的数据格式化为可被pytorch快速处理的张量类型
mnist_train = torchvision.datasets.FashionMNIST(
    root="../data", train=True, transform=trans, download=True)
mnist_test = torchvision.datasets.FashionMNIST(
    root="../data", train=False, transform=trans, download=True)

Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-images-idx3-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-images-idx3-ubyte.gz to ../data\FashionMNIST\raw\train-images-idx3-ubyte.gz


100.0%


Extracting ../data\FashionMNIST\raw\train-images-idx3-ubyte.gz to ../data\FashionMNIST\raw

Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-labels-idx1-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-labels-idx1-ubyte.gz to ../data\FashionMNIST\raw\train-labels-idx1-ubyte.gz


100.0%


Extracting ../data\FashionMNIST\raw\train-labels-idx1-ubyte.gz to ../data\FashionMNIST\raw

Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz to ../data\FashionMNIST\raw\t10k-images-idx3-ubyte.gz


100.0%


Extracting ../data\FashionMNIST\raw\t10k-images-idx3-ubyte.gz to ../data\FashionMNIST\raw

Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-labels-idx1-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-labels-idx1-ubyte.gz to ../data\FashionMNIST\raw\t10k-labels-idx1-ubyte.gz


100.0%

Extracting ../data\FashionMNIST\raw\t10k-labels-idx1-ubyte.gz to ../data\FashionMNIST\raw

​

​

'''
Fashion-MNIST 由 10 个类别的图像组成，每个类别由训练集的 6000 张图像和测试集的 1000 张图像组成，因此训练集和测试集长度分别为 60000 和 10000。
数据集由灰度图像组成，其通道数为1，每个图像的高度和宽度均为 28 像素。
'''
len(mnist_train), len(mnist_test), mnist_train[0][0].shape

(60000, 10000, torch.Size([1, 28, 28]))

def get_fashion_mnist_labels(labels):  #@save
    """返回Fashion-MNIST数据集的文本标签"""
    text_labels = ['t-shirt', 'trouser', 'pullover', 'dress', 'coat',
                   'sandal', 'shirt', 'sneaker', 'bag', 'ankle boot']  # Fashion-MNIST 中的 10 个类别
    return [text_labels[int(i)] for i in labels]

def show_images(imgs, num_rows, num_cols, titles=None, scale=1.5):  #@save
    """绘制图像列表"""
    figsize = (num_cols * scale, num_rows * scale)
    _, axes = d2l.plt.subplots(num_rows, num_cols, figsize=figsize)
    axes = axes.flatten()
    for i, (ax, img) in enumerate(zip(axes, imgs)):
        if torch.is_tensor(img):
            # 图片张量
            ax.imshow(img.numpy())
        else:
            # PIL图片
            ax.imshow(img)
        ax.axes.get_xaxis().set_visible(False)
        ax.axes.get_yaxis().set_visible(False)
        if titles:
            ax.set_title(titles[i])
    return axes

X, y = next(iter(data.DataLoader(mnist_train, batch_size=18)))
show_images(X.reshape(18, 28, 28), 2, 9, titles=get_fashion_mnist_labels(y))  # 演示绘制训练集前几个样本的图像及标签

array([<AxesSubplot:title={'center':'ankle boot'}>,
       <AxesSubplot:title={'center':'t-shirt'}>,
       <AxesSubplot:title={'center':'t-shirt'}>,
       <AxesSubplot:title={'center':'dress'}>,
       <AxesSubplot:title={'center':'t-shirt'}>,
       <AxesSubplot:title={'center':'pullover'}>,
       <AxesSubplot:title={'center':'sneaker'}>,
       <AxesSubplot:title={'center':'pullover'}>,
       <AxesSubplot:title={'center':'sandal'}>,
       <AxesSubplot:title={'center':'sandal'}>,
       <AxesSubplot:title={'center':'t-shirt'}>,
       <AxesSubplot:title={'center':'ankle boot'}>,
       <AxesSubplot:title={'center':'sandal'}>,
       <AxesSubplot:title={'center':'sandal'}>,
       <AxesSubplot:title={'center':'sneaker'}>,
       <AxesSubplot:title={'center':'ankle boot'}>,
       <AxesSubplot:title={'center':'trouser'}>,
       <AxesSubplot:title={'center':'t-shirt'}>], dtype=object)

​

​

3.5.2 读取小批量

batch_size = 256

def get_dataloader_workers():  #@save
    """使用4个进程来读取数据"""
    return 4

train_iter = data.DataLoader(mnist_train, batch_size, shuffle=True,  # 随机打乱
                             num_workers=get_dataloader_workers())

timer = d2l.Timer()
for X, y in train_iter:
    continue
f'{timer.stop():.2f}sec'  # 查看读取训练集所需的时间

'3.74sec'

3.5.3 整合所有组件

def load_data_fashion_mnist(batch_size, resize=None):  #@save  # resize 可用于指定要调整的图像尺寸
    """下载Fashion-MNIST数据集，然后将其加载到内存中"""
    trans = [transforms.ToTensor()]
    if resize:
        trans.insert(0, transforms.Resize(resize))
    trans = transforms.Compose(trans)
    mnist_train = torchvision.datasets.FashionMNIST(
        root="../data", train=True, transform=trans, download=True)
    mnist_test = torchvision.datasets.FashionMNIST(
        root="../data", train=False, transform=trans, download=True)
    return (data.DataLoader(mnist_train, batch_size, shuffle=True,
                            num_workers=get_dataloader_workers()),
            data.DataLoader(mnist_test, batch_size, shuffle=False,
                            num_workers=get_dataloader_workers()))

train_iter, test_iter = load_data_fashion_mnist(32, resize=64)
for X, y in train_iter:
    print(X.shape, X.dtype, y.shape, y.dtype)
    break

torch.Size([32, 1, 64, 64]) torch.float32 torch.Size([32]) torch.int64

练习

（1）减小 batch_size（如减小到1）是否会影响读取性能？

train_iter = data.DataLoader(mnist_train, 1, shuffle=True,  # batch_size 换成 1
                             num_workers=get_dataloader_workers())

timer = d2l.Timer()
for X, y in train_iter:
    continue
f'{timer.stop():.2f}sec'  # 读取性能大幅下降

'22.18sec'

---

（2）数据迭代器的性能非常重要。当前的实现足够快吗？探索各种选择来改进。

torch.utils.data.DataLoader 有个 pin_memory 参数，设置为 True 则开启内存固定，可加快加载速度。

详细参见文档 TORCH.UTILS.DATA

---

（3）查阅框架的在线 API 文档，还有哪些其他数据集可用？

详见文档 DATASETS