deeplabv3+源码之慢慢解析 第四章network文件夹(1)backbone文件夹(a1)hrnetv2.py--4个函数和可执行代码

**第四章 Network 文件夹**

在 DeeplabV3+ 的源码中，Network 文件夹包含了网络的定义和实现。我们将逐步解析这个文件夹。

**4.1 Backbone 文件夹**

Backbone 文件夹包含了网络的骨架结构，即特征提取部分。在这个文件夹下，我们找到一个名为 hrnetv2.py 的文件，这是 HRNetV2 网络的实现。

**hrnetv2.py**

import torchimport torch.nn as nnclass HRNetV2(nn.Module):
 def __init__(self, num_classes=21):
 super(HRNetV2, self).__init__()
 self.conv1 = nn.Conv2d(3,64, kernel_size=3)
 self.bn1 = nn.BatchNorm2d(64)
 self.relu1 = nn.ReLU()
 # Stage1 self.stage1 = nn.Sequential(
 nn.Conv2d(64,128, kernel_size=3),
 nn.BatchNorm2d(128),
 nn.ReLU(),
 nn.MaxPool2d(kernel_size=2, stride=2)
 )
 # Stage2 self.stage2 = nn.Sequential(
 nn.Conv2d(128,256, kernel_size=3),
 nn.BatchNorm2d(256),
 nn.ReLU(),
 nn.MaxPool2d(kernel_size=2, stride=2)
 )
 # Stage3 self.stage3 = nn.Sequential(
 nn.Conv2d(256,512, kernel_size=3),
 nn.BatchNorm2d(512),
 nn.ReLU(),
 nn.MaxPool2d(kernel_size=2, stride=2)
 )
 # Stage4 self.stage4 = nn.Sequential(
 nn.Conv2d(512,1024, kernel_size=3),
 nn.BatchNorm2d(1024),
 nn.ReLU()
 )
 # Decoder self.decoder = nn.Sequential(
 nn.Upsample(scale_factor=2),
 nn.Conv2d(1024,512, kernel_size=3),
 nn.BatchNorm2d(512),
 nn.ReLU(),
 nn.Upsample(scale_factor=2),
 nn.Conv2d(512,256, kernel_size=3),
 nn.BatchNorm2d(256),
 nn.ReLU(),
 nn.Upsample(scale_factor=2),
 nn.Conv2d(256, num_classes, kernel_size=3)
 )
 def forward(self, x):
 out = self.conv1(x)
 out = self.bn1(out)
 out = self.relu1(out)
 out = self.stage1(out)
 out = self.stage2(out)
 out = self.stage3(out)
 out = self.stage4(out)
 out = self.decoder(out)
 return out

model = HRNetV2(num_classes=21)
x = torch.randn(1,3,512,512) # 输入数据output = model(x)
print(output.shape)