输入一张图像，经过两次卷积后卷积后的输出特征图Pytorch

import torch

import torch.nn as nn

import matplotlib.pyplot as plt

import cv2

import torch.nn.functional as F

img = cv2.imread('E:/zhangzijiao/tu/29_29.png')#读入的图像是numpy形式

#然后对输入图像转换为tensor形式

img1=torch.tensor(img,dtype=torch.float32)/255

print(img1.shape) #此时图片的尺寸为（H,W,C）形式

img1=img1.permute(2,0,1).unsqueeze(0) #输入卷积中也必须有batch_size这一维度，permute(2,0,1)进行维度互换（C,H,W）

# 后一部分unsqueeze(0)是加入batch_size这一维度

print('1',img1.shape)#（B，C,H,W）

#conv=nn.Conv2d(in_channels=3,out_channels=3,kernel_size=3,stride=2,padding=1)

model = nn.Sequential(

nn.Conv2d(in_channels=3,out_channels=3,kernel_size=3,stride=1,padding=1,bias=False),

nn.Conv2d(in_channels=3,out_channels=3,kernel_size=3,stride=2,padding=1,bias=False),

nn.Conv2d(in_channels=3,out_channels=3,kernel_size=3,stride=2,padding=1,bias=False)

)

#print(conv)

y=model(img1)

print('2',y.shape)

model2 = nn.Sequential(nn.ConvTranspose2d(in_channels=3,out_channels=3,kernel_size=3,stride=2,padding=1,bias=False),

nn.ConvTranspose2d(in_channels=3,out_channels=3,kernel_size=3,stride=2,padding=1,bias=False),

)

#下边显示经过一次卷积后的图像

y = y.squeeze(0).permute(1,2,0)*255 #转化维度，并且转化为0~255，数值类型为uint8

y = y.detach().numpy() #tensor变成numpy

y = y.astype('uint8')

print('3',y.shape)

#权重默认随机可视化

plt.subplot(121)

plt.imshow(img)

plt.title('yuantu')

plt.subplot(122)

plt.imshow(y)

plt.title('liangcijuanji')

plt.show()