intermediate/spatial_transformer_tutorial
在 Google Colab 中运行
Colab
下载 Notebook
Notebook
在 GitHub 上查看
GitHub
注意
转到末尾下载完整的示例代码。
空间变换器网络教程#
创建于：2017年11月08日 | 最后更新：2024年01月19日 | 最后验证：2024年11月05日
在本教程中，您将学习如何使用一种称为空间变换器网络的视觉注意力机制来增强您的网络。您可以在 DeepMind 论文中阅读更多关于空间变换器网络的内容。
空间变换器网络是可微分注意力对任何空间变换的推广。空间变换器网络（简称STN）允许神经网络学习如何对输入图像执行空间变换，以增强模型的几何不变性。例如，它可以裁剪感兴趣的区域，缩放和校正图像的方向。这是一个有用的机制，因为CNN对旋转、缩放和更一般的仿射变换不具有不变性。
STN 的一大优点是能够以极少的修改轻松地将其集成到任何现有的 CNN 中。
# License: BSD
# Author: Ghassen Hamrouni
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 mode
<contextlib.ExitStack object at 0x7f5c66e606d0>
加载数据#
在这篇文章中，我们使用经典的 MNIST 数据集进行实验。使用一个标准的卷积网络，并用空间变换器网络对其进行增强。
from six.moves import urllib
opener = urllib.request.build_opener()
opener.addheaders = [('User-agent', 'Mozilla/5.0')]
urllib.request.install_opener(opener)
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)
0%| | 0.00/9.91M [00:00<?, ?B/s]
100%|██████████| 9.91M/9.91M [00:00<00:00, 139MB/s]
0%| | 0.00/28.9k [00:00<?, ?B/s]
100%|██████████| 28.9k/28.9k [00:00<00:00, 22.8MB/s]
0%| | 0.00/1.65M [00:00<?, ?B/s]
100%|██████████| 1.65M/1.65M [00:00<00:00, 253MB/s]
0%| | 0.00/4.54k [00:00<?, ?B/s]
100%|██████████| 4.54k/4.54k [00:00<00:00, 26.5MB/s]
描述空间变换器网络#
空间变换器网络可归结为三个主要部分
定位网络是一个常规的 CNN，它回归变换参数。变换不是从该数据集中明确学习的，而是网络自动学习能够提升全局精度的空间变换。
网格生成器在输入图像中生成一个坐标网格，对应于输出图像的每个像素。
采样器使用变换的参数，并将其应用于输入图像。
注意
我们需要最新版本的 PyTorch，其中包含 affine_grid 和 grid_sample 模块。
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)
训练模型#
现在，让我们使用 SGD 算法来训练模型。网络以监督的方式学习分类任务。同时，模型以端到端的方式自动学习 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 the STN 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)))
可视化 STN 结果#
现在，我们将检查我们学习到的视觉注意力机制的结果。
我们定义了一个小的辅助函数，以便在训练过程中可视化变换。
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()
/usr/local/lib/python3.10/dist-packages/torch/nn/functional.py:5163: UserWarning:
Default grid_sample and affine_grid behavior has changed to align_corners=False since 1.3.0. Please specify align_corners=True if the old behavior is desired. See the documentation of grid_sample for details.
/usr/local/lib/python3.10/dist-packages/torch/nn/functional.py:5096: UserWarning:
Default grid_sample and affine_grid behavior has changed to align_corners=False since 1.3.0. Please specify align_corners=True if the old behavior is desired. See the documentation of grid_sample for details.
Train Epoch: 1 [0/60000 (0%)] Loss: 2.373959
Train Epoch: 1 [32000/60000 (53%)] Loss: 0.698139
/usr/local/lib/python3.10/dist-packages/torch/nn/_reduction.py:51: UserWarning:
size_average and reduce args will be deprecated, please use reduction='sum' instead.
Test set: Average loss: 0.2587, Accuracy: 9283/10000 (93%)
Train Epoch: 2 [0/60000 (0%)] Loss: 0.469909
Train Epoch: 2 [32000/60000 (53%)] Loss: 0.543193
Test set: Average loss: 0.2270, Accuracy: 9342/10000 (93%)
Train Epoch: 3 [0/60000 (0%)] Loss: 0.357367
Train Epoch: 3 [32000/60000 (53%)] Loss: 0.361884
Test set: Average loss: 0.5571, Accuracy: 8700/10000 (87%)
Train Epoch: 4 [0/60000 (0%)] Loss: 0.936480
Train Epoch: 4 [32000/60000 (53%)] Loss: 0.193620
Test set: Average loss: 0.1455, Accuracy: 9590/10000 (96%)
Train Epoch: 5 [0/60000 (0%)] Loss: 0.254690
Train Epoch: 5 [32000/60000 (53%)] Loss: 0.200437
Test set: Average loss: 0.0694, Accuracy: 9809/10000 (98%)
Train Epoch: 6 [0/60000 (0%)] Loss: 0.256078
Train Epoch: 6 [32000/60000 (53%)] Loss: 0.107387
Test set: Average loss: 0.0801, Accuracy: 9754/10000 (98%)
Train Epoch: 7 [0/60000 (0%)] Loss: 0.301764
Train Epoch: 7 [32000/60000 (53%)] Loss: 0.073661
Test set: Average loss: 0.0632, Accuracy: 9817/10000 (98%)
Train Epoch: 8 [0/60000 (0%)] Loss: 0.413389
Train Epoch: 8 [32000/60000 (53%)] Loss: 0.156303
Test set: Average loss: 0.0937, Accuracy: 9728/10000 (97%)
Train Epoch: 9 [0/60000 (0%)] Loss: 0.141839
Train Epoch: 9 [32000/60000 (53%)] Loss: 0.149264
Test set: Average loss: 0.0608, Accuracy: 9822/10000 (98%)
Train Epoch: 10 [0/60000 (0%)] Loss: 0.079114
Train Epoch: 10 [32000/60000 (53%)] Loss: 0.111808
Test set: Average loss: 0.9517, Accuracy: 7849/10000 (78%)
Train Epoch: 11 [0/60000 (0%)] Loss: 1.049073
Train Epoch: 11 [32000/60000 (53%)] Loss: 0.151848
Test set: Average loss: 0.0497, Accuracy: 9838/10000 (98%)
Train Epoch: 12 [0/60000 (0%)] Loss: 0.133549
Train Epoch: 12 [32000/60000 (53%)] Loss: 0.076766
Test set: Average loss: 0.0493, Accuracy: 9853/10000 (99%)
Train Epoch: 13 [0/60000 (0%)] Loss: 0.081492
Train Epoch: 13 [32000/60000 (53%)] Loss: 0.050089
Test set: Average loss: 0.0488, Accuracy: 9856/10000 (99%)
Train Epoch: 14 [0/60000 (0%)] Loss: 0.081007
Train Epoch: 14 [32000/60000 (53%)] Loss: 0.139731
Test set: Average loss: 0.0488, Accuracy: 9860/10000 (99%)
Train Epoch: 15 [0/60000 (0%)] Loss: 0.052301
Train Epoch: 15 [32000/60000 (53%)] Loss: 0.097698
Test set: Average loss: 0.0496, Accuracy: 9860/10000 (99%)
Train Epoch: 16 [0/60000 (0%)] Loss: 0.074287
Train Epoch: 16 [32000/60000 (53%)] Loss: 0.247831
Test set: Average loss: 0.0493, Accuracy: 9858/10000 (99%)
Train Epoch: 17 [0/60000 (0%)] Loss: 0.118478
Train Epoch: 17 [32000/60000 (53%)] Loss: 0.008125
Test set: Average loss: 0.0439, Accuracy: 9872/10000 (99%)
Train Epoch: 18 [0/60000 (0%)] Loss: 0.113921
Train Epoch: 18 [32000/60000 (53%)] Loss: 0.087375
Test set: Average loss: 0.0677, Accuracy: 9811/10000 (98%)
Train Epoch: 19 [0/60000 (0%)] Loss: 0.121832
Train Epoch: 19 [32000/60000 (53%)] Loss: 0.139855
Test set: Average loss: 0.0419, Accuracy: 9876/10000 (99%)
Train Epoch: 20 [0/60000 (0%)] Loss: 0.037878
Train Epoch: 20 [32000/60000 (53%)] Loss: 0.059398
Test set: Average loss: 0.0471, Accuracy: 9859/10000 (99%)
脚本总运行时间： (1 分 37.715 秒)
