PyTorch实现ResNet50、ResNet101和ResNet152示例

PyTorch: https://github.com/shanglianlm0525/PyTorch-Networks

import torch
import torch.nn as nn
import torchvision
import numpy as np

print("PyTorch Version: ",torch.__version__)
print("Torchvision Version: ",torchvision.__version__)

__all__ = ['ResNet50', 'ResNet101','ResNet152']

def Conv1(in_planes, places, stride=2):
 return nn.Sequential(
   nn.Conv2d(in_channels=in_planes,out_channels=places,kernel_size=7,stride=stride,padding=3, bias=False),
   nn.BatchNorm2d(places),
   nn.ReLU(inplace=True),
   nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
 )

class Bottleneck(nn.Module):
 def __init__(self,in_places,places, stride=1,downsampling=False, expansion = 4):
   super(Bottleneck,self).__init__()
   self.expansion = expansion
   self.downsampling = downsampling

self.bottleneck = nn.Sequential(
     nn.Conv2d(in_channels=in_places,out_channels=places,kernel_size=1,stride=1, bias=False),
     nn.BatchNorm2d(places),
     nn.ReLU(inplace=True),
     nn.Conv2d(in_channels=places, out_channels=places, kernel_size=3, stride=stride, padding=1, bias=False),
     nn.BatchNorm2d(places),
     nn.ReLU(inplace=True),
     nn.Conv2d(in_channels=places, out_channels=places*self.expansion, kernel_size=1, stride=1, bias=False),
     nn.BatchNorm2d(places*self.expansion),
   )

if self.downsampling:
     self.downsample = nn.Sequential(
       nn.Conv2d(in_channels=in_places, out_channels=places*self.expansion, kernel_size=1, stride=stride, bias=False),
       nn.BatchNorm2d(places*self.expansion)
     )
   self.relu = nn.ReLU(inplace=True)
 def forward(self, x):
   residual = x
   out = self.bottleneck(x)

if self.downsampling:
     residual = self.downsample(x)

out += residual
   out = self.relu(out)
   return out

class ResNet(nn.Module):
 def __init__(self,blocks, num_classes=1000, expansion = 4):
   super(ResNet,self).__init__()
   self.expansion = expansion

self.conv1 = Conv1(in_planes = 3, places= 64)

self.layer1 = self.make_layer(in_places = 64, places= 64, block=blocks[0], stride=1)
   self.layer2 = self.make_layer(in_places = 256,places=128, block=blocks[1], stride=2)
   self.layer3 = self.make_layer(in_places=512,places=256, block=blocks[2], stride=2)
   self.layer4 = self.make_layer(in_places=1024,places=512, block=blocks[3], stride=2)

self.avgpool = nn.AvgPool2d(7, stride=1)
   self.fc = nn.Linear(2048,num_classes)

for m in self.modules():
     if isinstance(m, nn.Conv2d):
       nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')
     elif isinstance(m, nn.BatchNorm2d):
       nn.init.constant_(m.weight, 1)
       nn.init.constant_(m.bias, 0)

def make_layer(self, in_places, places, block, stride):
   layers = []
   layers.append(Bottleneck(in_places, places,stride, downsampling =True))
   for i in range(1, block):
     layers.append(Bottleneck(places*self.expansion, places))

return nn.Sequential(*layers)

def forward(self, x):
   x = self.conv1(x)

x = self.layer1(x)
   x = self.layer2(x)
   x = self.layer3(x)
   x = self.layer4(x)

x = self.avgpool(x)
   x = x.view(x.size(0), -1)
   x = self.fc(x)
   return x

def ResNet50():
 return ResNet([3, 4, 6, 3])

def ResNet101():
 return ResNet([3, 4, 23, 3])

def ResNet152():
 return ResNet([3, 8, 36, 3])

if __name__=='__main__':
 #model = torchvision.models.resnet50()
 model = ResNet50()
 print(model)

input = torch.randn(1, 3, 224, 224)
 out = model(input)
 print(out.shape)

来源：https://blog.csdn.net/shanglianlm/article/details/86376627