在本教程中，您將學習如何使用稱為空間變換器網絡的視覺注意力機制來擴充網絡。您可以在 DeepMind 論文中詳細了解空間變壓器網絡。
空間變換器網絡是對任何空間變換的可區別關注的概括。空間變換器網絡(簡稱 STN）允許神經網絡學習如何對輸入圖像執行空間變換，以增強模型的幾何不變性。例如，它可以裁剪感興趣的區域，縮放並校正圖像的方向。這可能是一個有用的機制，因為 CNN 不會對旋轉和縮放以及更一般的仿射變換保持不變。
關於 STN 的最好的事情之一就是能夠將它簡單地插入到任何現有的 CNN 中。
# License: BSD
# Author: Ghassen Hamrouni
from __future__ import print_function
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
import torchvision
from torchvision import datasets, transforms
import matplotlib.pyplot as plt
import numpy as np
plt.ion() # interactive modeCopy
在本文中，我們將嘗試使用經典的 MNIST 數據集。使用標準卷積網絡和空間變換器網絡。
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
# Training dataset
train_loader = torch.utils.data.DataLoader(
datasets.MNIST(root='.', train=True, download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,))
])), batch_size=64, shuffle=True, num_workers=4)
# Test dataset
test_loader = torch.utils.data.DataLoader(
datasets.MNIST(root='.', train=False, transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,))
])), batch_size=64, shuffle=True, num_workers=4)Copy
出：
Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz to ./MNIST/raw/train-images-idx3-ubyte.gz
Extracting ./MNIST/raw/train-images-idx3-ubyte.gz to ./MNIST/raw
Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz to ./MNIST/raw/train-labels-idx1-ubyte.gz
Extracting ./MNIST/raw/train-labels-idx1-ubyte.gz to ./MNIST/raw
Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz to ./MNIST/raw/t10k-images-idx3-ubyte.gz
Extracting ./MNIST/raw/t10k-images-idx3-ubyte.gz to ./MNIST/raw
Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz to ./MNIST/raw/t10k-labels-idx1-ubyte.gz
Extracting ./MNIST/raw/t10k-labels-idx1-ubyte.gz to ./MNIST/raw
Processing...
Done!Copy
空間變壓器網絡可歸結為三個主要組成部分：
Note
我們需要包含 affine_grid 和 grid_sample 模塊的最新版本的 PyTorch。
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
self.conv1 = nn.Conv2d(1, 10, kernel_size=5)
self.conv2 = nn.Conv2d(10, 20, kernel_size=5)
self.conv2_drop = nn.Dropout2d()
self.fc1 = nn.Linear(320, 50)
self.fc2 = nn.Linear(50, 10)
# Spatial transformer localization-network
self.localization = nn.Sequential(
nn.Conv2d(1, 8, kernel_size=7),
nn.MaxPool2d(2, stride=2),
nn.ReLU(True),
nn.Conv2d(8, 10, kernel_size=5),
nn.MaxPool2d(2, stride=2),
nn.ReLU(True)
)
# Regressor for the 3 * 2 affine matrix
self.fc_loc = nn.Sequential(
nn.Linear(10 * 3 * 3, 32),
nn.ReLU(True),
nn.Linear(32, 3 * 2)
)
# Initialize the weights/bias with identity transformation
self.fc_loc[2].weight.data.zero_()
self.fc_loc[2].bias.data.copy_(torch.tensor([1, 0, 0, 0, 1, 0], dtype=torch.float))
# Spatial transformer network forward function
def stn(self, x):
xs = self.localization(x)
xs = xs.view(-1, 10 * 3 * 3)
theta = self.fc_loc(xs)
theta = theta.view(-1, 2, 3)
grid = F.affine_grid(theta, x.size())
x = F.grid_sample(x, grid)
return x
def forward(self, x):
# transform the input
x = self.stn(x)
# Perform the usual forward pass
x = F.relu(F.max_pool2d(self.conv1(x), 2))
x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), 2))
x = x.view(-1, 320)
x = F.relu(self.fc1(x))
x = F.dropout(x, training=self.training)
x = self.fc2(x)
return F.log_softmax(x, dim=1)
model = Net().to(device)Copy
現在，讓我們使用 SGD 算法訓練模型。網絡正在以監督方式學習分類任務。同時，該模型以端到端的方式自動學習 STN。
optimizer = optim.SGD(model.parameters(), lr=0.01)
def train(epoch):
model.train()
for batch_idx, (data, target) in enumerate(train_loader):
data, target = data.to(device), target.to(device)
optimizer.zero_grad()
output = model(data)
loss = F.nll_loss(output, target)
loss.backward()
optimizer.step()
if batch_idx % 500 == 0:
print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
epoch, batch_idx * len(data), len(train_loader.dataset),
100\. * batch_idx / len(train_loader), loss.item()))
#
# A simple test procedure to measure STN the performances on MNIST.
#
def test():
with torch.no_grad():
model.eval()
test_loss = 0
correct = 0
for data, target in test_loader:
data, target = data.to(device), target.to(device)
output = model(data)
# sum up batch loss
test_loss += F.nll_loss(output, target, size_average=False).item()
# get the index of the max log-probability
pred = output.max(1, keepdim=True)[1]
correct += pred.eq(target.view_as(pred)).sum().item()
test_loss /= len(test_loader.dataset)
print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'
.format(test_loss, correct, len(test_loader.dataset),
100\. * correct / len(test_loader.dataset)))Copy
現在，我們將檢查學習到的視覺注意力機制的結果。
我們定義了一個小的輔助函數，以便在訓練時可視化轉換。
def convert_image_np(inp):
"""Convert a Tensor to numpy image."""
inp = inp.numpy().transpose((1, 2, 0))
mean = np.array([0.485, 0.456, 0.406])
std = np.array([0.229, 0.224, 0.225])
inp = std * inp + mean
inp = np.clip(inp, 0, 1)
return inp
# We want to visualize the output of the spatial transformers layer
# after the training, we visualize a batch of input images and
# the corresponding transformed batch using STN.
def visualize_stn():
with torch.no_grad():
# Get a batch of training data
data = next(iter(test_loader))[0].to(device)
input_tensor = data.cpu()
transformed_input_tensor = model.stn(data).cpu()
in_grid = convert_image_np(
torchvision.utils.make_grid(input_tensor))
out_grid = convert_image_np(
torchvision.utils.make_grid(transformed_input_tensor))
# Plot the results side-by-side
f, axarr = plt.subplots(1, 2)
axarr[0].imshow(in_grid)
axarr[0].set_title('Dataset Images')
axarr[1].imshow(out_grid)
axarr[1].set_title('Transformed Images')
for epoch in range(1, 20 + 1):
train(epoch)
test()
# Visualize the STN transformation on some input batch
visualize_stn()
plt.ioff()
plt.show()Copy
Out:
Train Epoch: 1 [0/60000 (0%)] Loss: 2.312544
Train Epoch: 1 [32000/60000 (53%)] Loss: 0.865688
Test set: Average loss: 0.2105, Accuracy: 9426/10000 (94%)
Train Epoch: 2 [0/60000 (0%)] Loss: 0.528199
Train Epoch: 2 [32000/60000 (53%)] Loss: 0.273284
Test set: Average loss: 0.1150, Accuracy: 9661/10000 (97%)
Train Epoch: 3 [0/60000 (0%)] Loss: 0.312562
Train Epoch: 3 [32000/60000 (53%)] Loss: 0.496166
Test set: Average loss: 0.1130, Accuracy: 9661/10000 (97%)
Train Epoch: 4 [0/60000 (0%)] Loss: 0.346181
Train Epoch: 4 [32000/60000 (53%)] Loss: 0.206084
Test set: Average loss: 0.0875, Accuracy: 9730/10000 (97%)
Train Epoch: 5 [0/60000 (0%)] Loss: 0.351175
Train Epoch: 5 [32000/60000 (53%)] Loss: 0.388225
Test set: Average loss: 0.0659, Accuracy: 9802/10000 (98%)
Train Epoch: 6 [0/60000 (0%)] Loss: 0.122667
Train Epoch: 6 [32000/60000 (53%)] Loss: 0.258372
Test set: Average loss: 0.0791, Accuracy: 9759/10000 (98%)
Train Epoch: 7 [0/60000 (0%)] Loss: 0.190197
Train Epoch: 7 [32000/60000 (53%)] Loss: 0.154468
Test set: Average loss: 0.0647, Accuracy: 9791/10000 (98%)
Train Epoch: 8 [0/60000 (0%)] Loss: 0.121149
Train Epoch: 8 [32000/60000 (53%)] Loss: 0.288490
Test set: Average loss: 0.0583, Accuracy: 9821/10000 (98%)
Train Epoch: 9 [0/60000 (0%)] Loss: 0.244609
Train Epoch: 9 [32000/60000 (53%)] Loss: 0.023396
Test set: Average loss: 0.0685, Accuracy: 9778/10000 (98%)
Train Epoch: 10 [0/60000 (0%)] Loss: 0.256878
Train Epoch: 10 [32000/60000 (53%)] Loss: 0.091626
Test set: Average loss: 0.0684, Accuracy: 9783/10000 (98%)
Train Epoch: 11 [0/60000 (0%)] Loss: 0.181910
Train Epoch: 11 [32000/60000 (53%)] Loss: 0.113193
Test set: Average loss: 0.0492, Accuracy: 9856/10000 (99%)
Train Epoch: 12 [0/60000 (0%)] Loss: 0.081072
Train Epoch: 12 [32000/60000 (53%)] Loss: 0.082513
Test set: Average loss: 0.0670, Accuracy: 9800/10000 (98%)
Train Epoch: 13 [0/60000 (0%)] Loss: 0.180748
Train Epoch: 13 [32000/60000 (53%)] Loss: 0.194512
Test set: Average loss: 0.0439, Accuracy: 9874/10000 (99%)
Train Epoch: 14 [0/60000 (0%)] Loss: 0.099560
Train Epoch: 14 [32000/60000 (53%)] Loss: 0.084377
Test set: Average loss: 0.0416, Accuracy: 9880/10000 (99%)
Train Epoch: 15 [0/60000 (0%)] Loss: 0.070021
Train Epoch: 15 [32000/60000 (53%)] Loss: 0.241336
Test set: Average loss: 0.0588, Accuracy: 9820/10000 (98%)
Train Epoch: 16 [0/60000 (0%)] Loss: 0.060536
Train Epoch: 16 [32000/60000 (53%)] Loss: 0.053016
Test set: Average loss: 0.0405, Accuracy: 9877/10000 (99%)
Train Epoch: 17 [0/60000 (0%)] Loss: 0.207369
Train Epoch: 17 [32000/60000 (53%)] Loss: 0.069607
Test set: Average loss: 0.1006, Accuracy: 9685/10000 (97%)
Train Epoch: 18 [0/60000 (0%)] Loss: 0.127503
Train Epoch: 18 [32000/60000 (53%)] Loss: 0.070724
Test set: Average loss: 0.0659, Accuracy: 9814/10000 (98%)
Train Epoch: 19 [0/60000 (0%)] Loss: 0.176861
Train Epoch: 19 [32000/60000 (53%)] Loss: 0.116980
Test set: Average loss: 0.0413, Accuracy: 9871/10000 (99%)
Train Epoch: 20 [0/60000 (0%)] Loss: 0.146933
Train Epoch: 20 [32000/60000 (53%)] Loss: 0.245741
Test set: Average loss: 0.0346, Accuracy: 9892/10000 (99%)