写在前面,代码白嫖时间长久,忘记是在哪里白嫖的了,emmmmm,如果有人发现出处,麻烦评论一下,我看见了就引用。
代码在colab平台测试可用,效果图也放在下面,只能保证代码可以用。
方法1
import matplotlib.pyplot as plt
import torch
from torchvision import models, transforms
from torch.utils.data import DataLoader, Dataset
import torch.nn as nn
from torch.autograd import Variable
import torch.nn.functional as F
import os
import numpy as np
from torchvision.datasets import ImageFolder
!unzip -d /content/ck /content/CK_Pro.zip
torch.cuda.set_device(0) # 设置GPU ID
is_cuda = True
simple_transform = transforms.Compose([transforms.Resize((224, 224)),
transforms.ToTensor(), # H, W, C -> C, W, H 归一化到(0,1),简单直接除以255
transforms.Normalize([0.485, 0.456, 0.406], # std
[0.229, 0.224, 0.225])])
# mean 先将输入归一化到(0,1),再使用公式”(x-mean)/std”,将每个元素分布到(-1,1)
# 使用 ImageFolder 必须有对应的目录结构
train = ImageFolder("/content/ck/CK_Pro/test", simple_transform)
# valid = ImageFolder("./datas/dogs-vs-cats/valid", simple_transform)
train_loader = DataLoader(train, batch_size=1, shuffle=False)
# val_loader = DataLoader(valid, batch_size=1, shuffle=False, num_workers=5)
vgg = models.vgg16(pretrained=True).cuda()
# 提取不同层输出的 主要代码
class LayerActivations:
features = None
def __init__(self, model, layer_num):
self.hook = model[layer_num].register_forward_hook(self.hook_fn)
def hook_fn(self, module, input, output):
self.features = output.cpu()
def remove(self):
self.hook.remove()
print(vgg.features)
conv_out = LayerActivations(vgg.features, 0) # 提出第 一个卷积层的输出
img = next(iter(train_loader))[0]
# imshow(img)
o = vgg(Variable(img.cuda()))
conv_out.remove() #
act = conv_out.features # act 即 第0层输出的特征
# 可视化 输出
fig = plt.figure()
# fig.subplots_adjust(left=0, right=0.5, bottom=0, top=0.8, hspace=0, wspace=0)
for i in range(64):
ax = fig.add_subplot(8, 8, i + 1, xticks=[], yticks=[])
ax.imshow(act[0][i].detach().numpy(), cmap="gray")
plt.show()
方法2
代码没跑出来,emmm,不想改代码