网站数据库安全人流什么时间做 新闻源网站

DAY 42 Grad-CAM与Hook函数

1.回调函数

def handle_result(result):print(f'计算结果是: {result}')def calculate(a, b, callback):result = a + bcallback(result)calculate(3, 5, handle_result)

计算结果是: 8

def handle_result(result):print(f'计算结果是: {result}')def with_callback(callback):def decorator(func):def wrapper(a, b):result = func(a, b)callback(result)return resultreturn wrapperreturn decorator@with_callback(handle_result)
def calculate(a, b):return a + bcalculate(3, 5)

计算结果是: 88

2.lambda函数

square = lambda x: x ** 2print(square(5))

25

3.hook函数的模块钩子和张量钩子

import torch
import torch.nn as nn
import numpy as np
import matplotlib.pyplot as plttorch.manual_seed(42)
np.random.seed(42)

import torch
import torch.nn as nnclass SimpleModel(nn.Module):def __init__(self):super(SimpleModel, self).__init__()self.conv = nn.Conv2d(1, 2, kernel_size=3, padding=1)self.relu = nn.ReLU()self.fc = nn.Linear(2 * 4 * 4, 10)def forward(self, x):x = self.conv(x)x = self.relu(x)x = x.view(-1, 2 * 4 * 4)x = self.fc(x)return xmodel = SimpleModel()
conv_outputs = []def forward_hook(module, input, output):print(f'钩子被调用！模块类型: {type(module)}')print(f'输入形状: {input[0].shape}')print(f'输出形状: {output.shape}')conv_outputs.append(output.detach())hook_handle = model.conv.register_forward_hook(forward_hook)
x = torch.randn(1, 1, 4, 4)
output = model(x)
hook_handle.remove()

钩子被调用！模块类型: <class 'torch.nn.modules.conv.Conv2d'>
输入形状: torch.Size([1, 1, 4, 4])
输出形状: torch.Size([1, 2, 4, 4])

import warnings
import matplotlib.pyplot as pltwarnings.filterwarnings('ignore')
plt.rcParams['font.family'] = ['SimHei']
plt.rcParams['axes.unicode_minus'] = Falseif conv_outputs:plt.figure(figsize=(10, 5))plt.subplot(1, 3, 1)plt.title('输入图像')plt.imshow(x[0, 0].detach().numpy(), cmap='gray')plt.subplot(1, 3, 2)plt.title('卷积核1输出')plt.imshow(conv_outputs[0][0, 0].detach().numpy(), cmap='gray')plt.subplot(1, 3, 3)plt.title('卷积核2输出')plt.imshow(conv_outputs[0][0, 1].detach().numpy(), cmap='gray')plt.tight_layout()plt.show()

conv_gradients = []def backward_hook(module, grad_input, grad_output):print(f'反向钩子被调用！模块类型: {type(module)}')print(f'输入梯度数量: {len(grad_input)}')print(f'输出梯度数量: {len(grad_output)}')conv_gradients.append((grad_input, grad_output))hook_handle = model.conv.register_backward_hook(backward_hook)
x = torch.randn(1, 1, 4, 4, requires_grad=True)
output = model(x)
loss = output.sum()
loss.backward()
hook_handle.remove()

反向钩子被调用！模块类型: <class 'torch.nn.modules.conv.Conv2d'>
输入梯度数量: 3
输出梯度数量: 1

x = torch.tensor([2.0], requires_grad=True)
y = x ** 2
z = y ** 3def tensor_hook(grad):print(f'原始梯度: {grad}')return grad / 2hook_handle = y.register_hook(tensor_hook)
z.backward()
print(f'x的梯度: {x.grad}')
hook_handle.remove()

原始梯度: tensor([48.])
x的梯度: tensor([96.])

4.Grad-CAM的示例

import torch
import torch.nn as nn
import torch.nn.functional as F
import torchvision
import torchvision.transforms as transforms
import numpy as np
import matplotlib.pyplot as plt
from PIL import Imagetorch.manual_seed(42)
np.random.seed(42)transform = transforms.Compose([transforms.ToTensor(),transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])testset = torchvision.datasets.CIFAR10(root='./data', train=False,download=True, transform=transform
)classes = ('飞机', '汽车', '鸟', '猫', '鹿', '狗', '青蛙', '马', '船', '卡车')class SimpleCNN(nn.Module):def __init__(self):super(SimpleCNN, self).__init__()self.conv1 = nn.Conv2d(3, 32, kernel_size=3, padding=1)self.conv2 = nn.Conv2d(32, 64, kernel_size=3, padding=1)self.conv3 = nn.Conv2d(64, 128, kernel_size=3, padding=1)self.pool = nn.MaxPool2d(2, 2)self.fc1 = nn.Linear(128 * 4 * 4, 512)self.fc2 = nn.Linear(512, 10)def forward(self, x):x = self.pool(F.relu(self.conv1(x)))x = self.pool(F.relu(self.conv2(x)))x = self.pool(F.relu(self.conv3(x)))x = x.view(-1, 128 * 4 * 4)x = F.relu(self.fc1(x))x = self.fc2(x)return xmodel = SimpleCNN()
print('模型已创建')
device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')
model = model.to(device)def train_model(model, epochs=1):trainset = torchvision.datasets.CIFAR10(root='./data', train=True,download=True, transform=transform)trainloader = torch.utils.data.DataLoader(trainset, batch_size=64,shuffle=True, num_workers=2)criterion = nn.CrossEntropyLoss()optimizer = torch.optim.Adam(model.parameters(), lr=0.001)for epoch in range(epochs):running_loss = 0.0for i, data in enumerate(trainloader, 0):inputs, labels = datainputs, labels = inputs.to(device), labels.to(device)optimizer.zero_grad()outputs = model(inputs)loss = criterion(outputs, labels)loss.backward()optimizer.step()running_loss += loss.item()if i % 100 == 99:print(f'[{epoch + 1}, {i + 1}] 损失: {running_loss / 100:.3f}')running_loss = 0.0print('训练完成')try:model.load_state_dict(torch.load('cifar10_cnn.pth'))print('已加载预训练模型')
except:print('无法加载预训练模型, 使用未训练模型或训练新模型')train_model(model, epochs=1)torch.save(model.state_dict(), 'cifar10_cnn.pth')model.eval()class GradCAM:def __init__(self, model, target_layer):self.model = modelself.target_layer = target_layerself.gradients = Noneself.activations = Noneself.register_hooks()def register_hooks(self):def forward_hook(module, input, output):self.activations = output.detach()def backward_hook(module, grad_input, grad_output):self.gradients = grad_output[0].detach()self.target_layer.register_forward_hook(forward_hook)self.target_layer.register_backward_hook(backward_hook)def generate_cam(self, input_image, target_class=None):model_output = self.model(input_image)if target_class is None:target_class = torch.argmax(model_output, dim=1).item()self.model.zero_grad()one_hot = torch.zeros_like(model_output)one_hot[0, target_class] = 1model_output.backward(gradient=one_hot)gradients = self.gradientsactivations = self.activationsweights = torch.mean(gradients, dim=(2, 3), keepdim=True)cam = torch.sum(weights * activations, dim=1, keepdim=True)cam = F.relu(cam)cam = F.interpolate(cam, size=(32, 32), mode='bilinear', align_corners=False)cam = cam - cam.min()cam = cam / cam.max() if cam.max() > 0 else camreturn cam.cpu().squeeze().numpy(), target_class

Files already downloaded and verified
模型已创建
已加载预训练模型

import warnings
import matplotlib.pyplot as pltwarnings.filterwarnings('ignore')
plt.rcParams['font.family'] = ['SimHei']
plt.rcParams['axes.unicode_minus'] = Falseidx = 102
image, label = testset[idx]
print(f'选择的图像类别: {classes[label]}')def tensor_to_np(tensor):img = tensor.cpu().numpy().transpose(1, 2, 0)mean = np.array([0.5, 0.5, 0.5])std = np.array([0.5, 0.5, 0.5])img = std * img + meanimg = np.clip(img, 0, 1)return imginput_tensor = image.unsqueeze(0).to(device)
grad_cam = GradCAM(model, model.conv3)
heatmap, pred_class = grad_cam.generate_cam(input_tensor)plt.figure(figsize=(12, 4))
plt.subplot(1, 3, 1)
plt.imshow(tensor_to_np(image))
plt.title(f'原始图像: {classes[label]}')
plt.axis('off')plt.subplot(1, 3, 2)
plt.imshow(heatmap, cmap='jet')
plt.title(f'Grad-CAM热力图: {classes[pred_class]}')
plt.axis('off')plt.subplot(1, 3, 3)
img = tensor_to_np(image)
heatmap_resized = np.uint8(255 * heatmap)
heatmap_colored = plt.cm.jet(heatmap_resized)[:, :, :3]
superimposed_img = heatmap_colored * 0.4 + img * 0.6
plt.imshow(superimposed_img)
plt.title('叠加热力图')
plt.axis('off')plt.tight_layout()
plt.savefig('grad_cam_result.png')
plt.show()

选择的图像类别: 青蛙

作业：理解下今天的代码即可

@浙大疏锦行