[Pytorch 기초 - 2] MNIST data를 load와 시각화 작업 구현하기

 

 

 

 

 

https://github.com/pytorch/examples/tree/master/mnist

 

PyTorch Data Preprocess¶

In [3]:

import torch

from torchvision import datasets, transforms

 

Data Loader 부르기¶

파이토치는 DataLoader를 불러 model에 넣음

In [4]:

batch_size = 32
test_batch_size = 32

In [10]:

train_loader = torch.utils.data.DataLoader(
    datasets.MNIST('dataset/',train = True,download = True,
                 transform = transforms.Compose([
                     transforms.ToTensor(),
                     transforms.Normalize(mean = (0.5,), std = (0.5,))
                 ])),
    batch_size = batch_size,
    shuffle = True)

 

Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz to dataset/MNIST\raw\train-images-idx3-ubyte.gz

 

 

 

Extracting dataset/MNIST\raw\train-images-idx3-ubyte.gz to dataset/MNIST\raw
Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz to dataset/MNIST\raw\train-labels-idx1-ubyte.gz

 

 

 

Extracting dataset/MNIST\raw\train-labels-idx1-ubyte.gz to dataset/MNIST\raw
Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz to dataset/MNIST\raw\t10k-images-idx3-ubyte.gz

 

 

 

Extracting dataset/MNIST\raw\t10k-images-idx3-ubyte.gz to dataset/MNIST\raw
Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz to dataset/MNIST\raw\t10k-labels-idx1-ubyte.gz

 

 

 

Extracting dataset/MNIST\raw\t10k-labels-idx1-ubyte.gz to dataset/MNIST\raw
Processing...
Done!

 

C:\Users\user\Anaconda3\lib\site-packages\torchvision\datasets\mnist.py:469: UserWarning: The given NumPy array is not writeable, and PyTorch does not support non-writeable tensors. This means you can write to the underlying (supposedly non-writeable) NumPy array using the tensor. You may want to copy the array to protect its data or make it writeable before converting it to a tensor. This type of warning will be suppressed for the rest of this program. (Triggered internally at  ..\torch\csrc\utils\tensor_numpy.cpp:141.)
  return torch.from_numpy(parsed.astype(m[2], copy=False)).view(*s)

In [12]:

test_loader = torch.utils.data.DataLoader(
    datasets.MNIST('dataset',train=False,
                  transform = transforms.Compose([
                      transforms.ToTensor(),
                      transforms.Normalize((0.5,), (0,5))
                  ])),
    batch_size = test_batch_size,
    shuffle = True)

 

첫번재 iteration에서 나오는 데이터 확인¶

In [13]:

images, labels = next(iter(train_loader))

In [17]:

images.shape 
# tensorflow에서는  (batch_size, height, width, channel) - (32, 28, 28, 1)
# pytorch에서는  (batch_size, channel, height, width) - (32, 1, 28, 28)
# rgb였으면 1이 아니라 3
# pytorch와 tensorflow의 차이점 !!

Out[17]:

torch.Size([32, 1, 28, 28])

In [18]:

labels.shape

Out[18]:

torch.Size([32])

 

PyTorch는 TensorFlow와 다르게 [Batch Size, Channel, Height, Width] 임을 명시해야함

 

데이터 시각화¶

In [19]:

import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline

In [20]:

torch_image = torch.squeeze(images[0])
torch_image.shape

Out[20]:

torch.Size([28, 28])

In [22]:

image = torch_image.numpy()
image.shape

Out[22]:

(28, 28)

In [23]:

label = labels[0].numpy()

In [24]:

label.shape

Out[24]:

()

In [25]:

label

Out[25]:

array(8, dtype=int64)

In [27]:

plt.title(label)
plt.imshow(image,'gray')
plt.show()