import torch
import torch.nn as nn
import torch.optim as optim
# 배치크기 * 채널(1: 그레이스케일, 3: 컬러, 4:컬러+투명도) * 너비 * 높이
inputs = torch.Tensor(1, 1, 28, 28)
print(inputs.shape)
torch.Size([1, 1, 28, 28])
# 첫번째 Conv2D
conv1 = nn.Conv2d(in_channels=1, out_channels=32, kernel_size=3, padding='same')
out = conv1(inputs)
print(out.shape)
torch.Size([1, 32, 28, 28])
# 첫번째 MaxPool2D
pool = nn.MaxPool2d(kernel_size=2)
out = pool(out)
print(out.shape)
torch.Size([1, 32, 14, 14])
# 두번째 Conv2D
conv2 = nn.Conv2d(in_channels=32, out_channels=64, kernel_size=3, padding='same')
out = conv2(out)
print(out.shape)
torch.Size([1, 64, 14, 14])
# 두번째 MaxPool2D
pool = nn.MaxPool2d(kernel_size=2)
out = pool(out)
print(out.shape)
torch.Size([1, 64, 7, 7])
flatten = nn.Flatten()
out = flatten(out)
print(out.shape) # 64 * 7 * 7
torch.Size([1, 3136])
fc = nn.Linear(3136, 10)
out = fc(out)
print(out.shape)
torch.Size([1, 10])
CNN으로 MNIST 분류하기¶
import torchvision.datasets as datasets
import torchvision.transforms as transforms
import matplotlib.pyplot as plt
from torch.utils.data import DataLoader
device = 'cuda' if torch.cuda.is_available() else 'cpu'
print(device)
cuda
train_data = datasets.MNIST(
root = 'data',
train = True,
transform = transforms.ToTensor(),
download = True
)
Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz
Failed to download (trying next):
HTTP Error 403: Forbidden
Downloading https://ossci-datasets.s3.amazonaws.com/mnist/train-images-idx3-ubyte.gz
Downloading https://ossci-datasets.s3.amazonaws.com/mnist/train-images-idx3-ubyte.gz to data/MNIST/raw/train-images-idx3-ubyte.gz
100%|██████████| 9912422/9912422 [00:01<00:00, 5435792.23it/s]
Extracting data/MNIST/raw/train-images-idx3-ubyte.gz to data/MNIST/raw
Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz
Failed to download (trying next):
HTTP Error 403: Forbidden
Downloading https://ossci-datasets.s3.amazonaws.com/mnist/train-labels-idx1-ubyte.gz
Downloading https://ossci-datasets.s3.amazonaws.com/mnist/train-labels-idx1-ubyte.gz to data/MNIST/raw/train-labels-idx1-ubyte.gz
100%|██████████| 28881/28881 [00:00<00:00, 156515.77it/s]
Extracting data/MNIST/raw/train-labels-idx1-ubyte.gz to data/MNIST/raw
Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz
Failed to download (trying next):
HTTP Error 403: Forbidden
Downloading https://ossci-datasets.s3.amazonaws.com/mnist/t10k-images-idx3-ubyte.gz
Downloading https://ossci-datasets.s3.amazonaws.com/mnist/t10k-images-idx3-ubyte.gz to data/MNIST/raw/t10k-images-idx3-ubyte.gz
100%|██████████| 1648877/1648877 [00:01<00:00, 1506435.10it/s]
Extracting data/MNIST/raw/t10k-images-idx3-ubyte.gz to data/MNIST/raw
Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz
Failed to download (trying next):
HTTP Error 403: Forbidden
Downloading https://ossci-datasets.s3.amazonaws.com/mnist/t10k-labels-idx1-ubyte.gz
Downloading https://ossci-datasets.s3.amazonaws.com/mnist/t10k-labels-idx1-ubyte.gz to data/MNIST/raw/t10k-labels-idx1-ubyte.gz
100%|██████████| 4542/4542 [00:00<00:00, 4630658.43it/s]Extracting data/MNIST/raw/t10k-labels-idx1-ubyte.gz to data/MNIST/raw
test_data = datasets.MNIST(
root = 'data',
train = False,
transform = transforms.ToTensor(),
download = True
)
print(train_data)
print(test_data)
Dataset MNIST
Number of datapoints: 60000
Root location: data
Split: Train
StandardTransform
Transform: ToTensor()
Dataset MNIST
Number of datapoints: 10000
Root location: data
Split: Test
StandardTransform
Transform: ToTensor()
loader = DataLoader(
dataset = train_data,
batch_size = 64,
shuffle = True
)
imgs, labels = next(iter(loader))
fig, axes = plt.subplots(8, 8, figsize=(16, 16))
for ax, img, label in zip(axes.flatten(), imgs, labels):
ax.imshow(img.reshape((28, 28)), cmap='gray')
ax.set_title(label.item())
ax.axis('off')
model = nn.Sequential(
nn.Conv2d(1, 32, kernel_size=3, padding='same'),
nn.ReLU(),
nn.MaxPool2d(kernel_size=2),
nn.Conv2d(32, 64, kernel_size=3, padding='same'),
nn.ReLU(),
nn.MaxPool2d(kernel_size=2),
nn.Flatten(),
nn.Linear(64 * 7 * 7, 10)
).to(device)
print(model)
Sequential(
(0): Conv2d(1, 32, kernel_size=(3, 3), stride=(1, 1), padding=same)
(1): ReLU()
(2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
(3): Conv2d(32, 64, kernel_size=(3, 3), stride=(1, 1), padding=same)
(4): ReLU()
(5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
(6): Flatten(start_dim=1, end_dim=-1)
(7): Linear(in_features=3136, out_features=10, bias=True)
)
optimizer = optim.Adam(model.parameters(), lr=0.001)
epochs = 10
for epoch in range(epochs):
sum_losses = 0
sum_accs = 0
for x_batch, y_batch in loader:
x_batch = x_batch.to(device)
y_batch = y_batch.to(device)
y_pred = model(x_batch)
loss = nn.CrossEntropyLoss()(y_pred, y_batch)
optimizer.zero_grad()
loss.backward()
optimizer.step()
sum_losses = sum_losses + loss
y_prob = nn.Softmax(1)(y_pred)
y_pred_index = torch.argmax(y_prob, axis=1)
acc = (y_batch == y_pred_index).float().sum() / len(y_batch) * 100
sum_accs = sum_accs + acc
avg_loss = sum_losses / len(loader)
avg_acc = sum_accs / len(loader)
print(f'Epoch {epoch:4d}/{epochs} Loss: {avg_loss:.6f} Accuracy: {avg_acc:.2f}%')
Epoch 0/10 Loss: 0.173020 Accuracy: 94.81%
Epoch 1/10 Loss: 0.055531 Accuracy: 98.23%
Epoch 2/10 Loss: 0.040530 Accuracy: 98.74%
Epoch 3/10 Loss: 0.031358 Accuracy: 99.04%
Epoch 4/10 Loss: 0.026460 Accuracy: 99.18%
Epoch 5/10 Loss: 0.021231 Accuracy: 99.32%
Epoch 6/10 Loss: 0.017277 Accuracy: 99.45%
Epoch 7/10 Loss: 0.013603 Accuracy: 99.57%
Epoch 8/10 Loss: 0.012516 Accuracy: 99.58%
Epoch 9/10 Loss: 0.010684 Accuracy: 99.63%
test_loader = DataLoader(
dataset = test_data,
batch_size = 64,
shuffle = True
)
imgs, labels = next(iter(test_loader))
fig, axes = plt.subplots(8, 8, figsize=(16, 16))
for ax, img, label in zip(axes.flatten(), imgs, labels):
ax.imshow(img.reshape((28, 28)), cmap='gray')
ax.set_title(label.item())
ax.axis('off')
model.eval() # 모델을 테스트 모드로 전환
sum_accs = 0
for x_batch, y_batch in test_loader:
x_batch = x_batch.to(device)
y_batch = y_batch.to(device)
y_pred = model(x_batch)
y_prob = nn.Softmax(1)(y_pred)
y_pred_index = torch.argmax(y_prob, axis=1)
acc = (y_batch == y_pred_index).float().sum() / len(y_batch) * 100
sum_accs = sum_accs + acc
avg_acc = sum_accs / len(test_loader)
print(f'테스트 정확도는 {avg_acc:.2f}%')