分类使用Pytorch实现Grad-CAM并绘制热力图

import os
import numpy as np
import torch
from PIL import Image
import matplotlib.pyplot as plt
from torchvision import models
from torchvision import transforms
import cv2
# from utils import GradCAM, show_cam_on_image, center_crop_img

    
activations  = []
gradients  = []
handles = []

def save_activation( module, input, output):
    activation = output
    activations.append(activation.cpu().detach())

def save_gradient(module, grad_input, grad_output):
    grad = grad_output[0]
    global gradients
    gradients = [grad.cpu().detach()] + gradients


def get_loss(output, target_category): # 这是为了batch size != 1的情况准备的
    loss = 0
    for i in range(len(target_category)):
        loss = loss + output[i, target_category[i]]
    return loss



def scale_cam_image(cam, target_size=None):
    result = []
    for img in cam:
        img = img - np.min(img)
        img = img / (1e-7 + np.max(img))
        if target_size is not None:
            img = cv2.resize(img, target_size)
        result.append(img)
    result = np.float32(result)

    return result
def aggregate_multi_layers( cam_per_target_layer):
        cam_per_target_layer = np.concatenate(cam_per_target_layer, axis=1)
        cam_per_target_layer = np.maximum(cam_per_target_layer, 0)
        result = np.mean(cam_per_target_layer, axis=1)
        return scale_cam_image(result)

def show_cam_on_image(img: np.ndarray,
                      mask: np.ndarray,
                      use_rgb: bool = False,
                      colormap: int = cv2.COLORMAP_JET) -> np.ndarray:
    """ This function overlays the cam mask on the image as an heatmap.
    By default the heatmap is in BGR format.
    :param img: The base image in RGB or BGR format.
    :param mask: The cam mask.
    :param use_rgb: Whether to use an RGB or BGR heatmap, this should be set to True if 'img' is in RGB format.
    :param colormap: The OpenCV colormap to be used.
    :returns: The default image with the cam overlay.
    """

    heatmap = cv2.applyColorMap(np.uint8(255 * mask), colormap)
    if use_rgb:
        heatmap = cv2.cvtColor(heatmap, cv2.COLOR_BGR2RGB)
    heatmap = np.float32(heatmap) / 255

    if np.max(img) > 1:
        raise Exception(
            "The input image should np.float32 in the range [0, 1]")

    cam = heatmap + img
    cam = cam / np.max(cam)
    return np.uint8(255 * cam)

def main():
    model = models.mobilenet_v3_large(pretrained=True)
    # 获取最分类层之前的
    target_layers = [model.features[-1]]


    data_transform =  transforms.Compose([transforms.ToTensor(),
                                         transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])

    img_path = "/data/ZA/medicine/remote_sensing_image/code/tensorboard/dog.jpg"
    assert os.path.exists(img_path), "file: '{}' dose not exist.".format(img_path)
    img = Image.open(img_path).convert('RGB')
    img = np.array(img, dtype=np.uint8)
    # img = cv2.reszie()
    print(img.shape)
    # [C, H, W]
    img_tensor = data_transform(img)
    # expand batch dimension
    # [C, H, W] -> [N, C, H, W]
    input_tensor = torch.unsqueeze(img_tensor, dim=0)
    # target_category = 3

    target_category = 254  # pug, pug-dog           
    for target_layer in target_layers:
        handles.append(
            target_layer.register_forward_hook(
                save_activation
            )
        )
        # Backward compatibility with older pytorch versions:
        if hasattr(target_layer, 'register_full_backward_hook'):
            handles.append(
                target_layer.register_full_backward_hook(
                    save_gradient))
        else:
            handles.append(
                target_layer.register_backward_hook(
                    save_gradient))



    print(handles[0])  # 在这 现在只是一个hook的状态 所以我们所以现在来看 没有进行反向传播的时候好像什么用都没有
    print(activations) # 这时候里面什么参数都没有

    # forward
    model.eval()
    output = model(input_tensor)
    #  # 正向传播得到网络输出logits(未经过softmax)

    target_category = [target_category] * input_tensor.size(0) # 将结果格式话
    model.zero_grad()                                           # 梯度归零
    loss = get_loss(output,target_category)                     # 计算损失
    loss.backward(retain_graph=True)                            # 反向传播

    # print(handles) 这个时候就有了参数了

    activations_list = [a.cpu().data.numpy()                                
                            for a in activations] 
    grads_list = [g.cpu().data.numpy()
                      for g in gradients]
    
    target_size  = input_tensor.size(-1), input_tensor.size(-2)
    cam_per_target_layer = []
    for layer_activations,layer_grads in zip(activations_list,grads_list):
        weights = np.mean(layer_grads, axis=(2, 3), keepdims=True)  # 使用均值作为权重
        # print(weights)
        weighted_activations = weights * layer_activations          # 对应相乘在相加
        cam = weighted_activations.sum(axis=1)
        cam[cam < 0] = 0  # relu
        scaled = scale_cam_image(cam, target_size)
        cam_per_target_layer.append(scaled[:, None, :])
    
    grayscale_cam = aggregate_multi_layers(cam_per_target_layer) # 将多层进行合并
    grayscale_cam = grayscale_cam[0]
    visualization = show_cam_on_image(img.astype(dtype=np.float32) / 255.,
                                      grayscale_cam,
                                      use_rgb=True)
    plt.imsave("./cam.png",visualization)



    # cam = GradCAM(model=model, target_layers=target_layers, use_cuda=False)
    # target_category = 281  # tabby, tabby cat
    # target_category = 254  # pug, pug-dog    

    # grayscale_cam = cam(input_tensor=input_tensor, target_category=target_category)

    # grayscale_cam = grayscale_cam[0, :]
    # visualization = show_cam_on_image(img.astype(dtype=np.float32) / 255.,
    #                                   grayscale_cam,
    #                                   use_rgb=True)
    # plt.imsave("./cam.png",visualization)



if __name__ == "__main__":
    main()

具体实现和细节部分 可以参考b站大神https://www.bilibili.com/video/BV1e3411j7x7?spm_id_from=333.999.0.0的视频 我也就是按照他的方法来实现的

其次就是我发现的一点小问题就是感觉没有eval 要好一点
请大知道的大佬教导一下

当我的反向传播是狗的时候

当我把eval模式退出之后：
我觉得相对合理了

当是猫的时候其实差别就不大了：
有eval:

无eval：