How to visualizing a sample
井民全, Jing, mqjing@gmail.com
This example is reference to the pytorch tutorial[1]. If you want to get detail information, check it.
Key
# get a sample tensor_img_1x28x28, label_idx = training_data[sample_idx] # show it figure = plt.figure(figsize=(1, 1)) i = 1 cols, rows = 1, 1 figure.add_subplot(rows, cols, i) plt.imshow(tensor_img_1x28x28.squeeze(), cmap="gray") plt.axis("off") plt.show() |
Code
File: test.py
import torch from torch.utils.data import Dataset from torchvision import datasets from torchvision.transforms import ToTensor import matplotlib.pyplot as plt # Step 1: Load datasets training_data = datasets.FashionMNIST( root="data", train=True, download=True, transform=ToTensor() ) test_data = datasets.FashionMNIST( root="data", train=False, download=True, transform=ToTensor() ) # Step 2: Get a sample with type tensor # randint: returns a tensor filled with random integers generated uniformly between 0 and high (exclusive) with shape = (1,) # that means 1d array with only 1 elements tensor_uniform_1x1 = torch.randint(len(training_data), size=(1,)) sample_idx = tensor_uniform_1x1.item() print('tensor_uniform = ', tensor_uniform_1x1, 'tensor.shape', tensor_uniform_1x1.shape, ' tensor_uniform_1x1.numpy().shape = ', tensor_uniform_1x1.numpy().shape) tensor_img_1x28x28, label_idx = training_data[sample_idx] print('tensor_img_1x28x28.shape = ', tensor_img_1x28x28.shape) # img_1x28x28.shape = torch.Size([1, 28, 28]) # Step 3: Squeeze the sample with sahpe [1, 28, 28] to the 2d image with shape [28, 28] tensor_img_28x28 = tensor_img_1x28x28.squeeze() print('tensor_img_28x28 = ', tensor_img_28x28, 'tensor_img_28x28.shape = ', tensor_img_28x28.shape) # torch.Size([28, 28]) figure = plt.figure(figsize=(1, 1)) i = 1 cols, rows = 1, 1 figure.add_subplot(rows, cols, i) plt.imshow(tensor_img_28x28, cmap="gray") plt.axis("off") plt.show() # figure = plt.figure(figsize=(8, 8)) # cols, rows = 3, 3 # for i in range(1, cols * rows + 1): # sample_idx = torch.randint(len(training_data), size=(1,)).item() # img, label = training_data[sample_idx] # figure.add_subplot(rows, cols, i) # plt.axis("off") # plt.imshow(img.squeeze(), cmap="gray") # plt.show() |
