Pytorch源码学习之六：torchvision.models.inception_v3

0. 参考和简介

torchvision.models.inception_v3源码
实现的论文地址
部分网络结构：

• InceptionA
  
• InceptionC
  
• InceptionE
  

一、源码

#网络模型:https://arxiv.org/abs/1512.00567
from collections import namedtuple
import torch
import torch.nn as nn
import  torch.nn.functional as F
from torch.hub import load_state_dict_from_url

__all__ = ['Inception3', 'inception_v3']

model_urls = {
    # Inception v3 ported from TensorFlow
    'inception_v3_google': 'https://download.pytorch.org/models/inception_v3_google-1a9a5a14.pth',
}

_InceptionOuputs = namedtuple('InceptionOuputs', ['logits', 'aux_logits'])

class BasicConv2d(nn.Module):
    def __init__(self, in_channels, out_channels, **kwargs):
        super(BasicConv2d, self).__init__()
        self.conv = nn.Conv2d(in_channels, out_channels, bias=False, **kwargs)
        self.bn = nn.BatchNorm2d(out_channels, eps=0.001)

    def forward(self, x):
        x = self.conv(x)
        x = self.bn (x)

        return F.relu(x, inplace=True)

class InceptionA(nn.Module):

    def __init__(self, in_channels, pool_features):
        super(InceptionA, self).__init__()
        self.branch1x1 = BasicConv2d(in_channels, 64, kernel_size=1)

        self.branch5x5_1 = BasicConv2d(in_channels, 48, kernel_size=1)
        self.branch5x5_2 = BasicConv2d(48, 64, kernel_size=5, padding=2)

        self.branch3x3dbl_1 = BasicConv2d(in_channels, 64, kernel_size=1)
        self.branch3x3dbl_2 = BasicConv2d(64, 96, kernel_size=3, padding=1)
        self.branch3x3dbl_3 = BasicConv2d(96, 96, kernel_size=3, padding=1)

        self.branch_pool = BasicConv2d(in_channels, pool_features, kernel_size=1)

    def forward(self, x):
        branch1x1 = self.branch1x1(x)

        branch5x5 = self.branch5x5_1(x)
        branch5x5 = self.branch5x5_2(branch5x5)

        branch3x3dbl = self.branch3x3dbl_1(x)
        branch3x3dbl = self.branch3x3dbl_2(branch3x3dbl)
        branch3x3dbl = self.branch3x3dbl_3(branch3x3dbl)

        branch_pool = F.avg_pool2d(x, kernel_size=3, stride=1, padding=1)
        branch_pool = self.branch_pool(branch_pool)

        outputs = [branch1x1, branch5x5, branch3x3dbl, branch_pool]
        return torch.cat(outputs, 1)

class InceptionB(nn.Module):

    def __init__(self, in_channels):
        super(InceptionB, self).__init__()
        self.branch3x3 = BasicConv2d(in_channels, 384, kernel_size=3, stride=2)

        self.brach3x3dbl_1 = BasicConv2d(in_channels, 64, kernel_size=1)
        self.brach3x3dbl_2 = BasicConv2d(64, 96, kernel_size=3, padding=1)
        self.brach3x3dbl_3 = BasicConv2d(96, 96, kernel_size=3, stride=2)

    def forward(self, x):
        branch3x3 = self.branch3x3(x)

        branch3x3dbl = self.brach3x3dbl_1(x)
        branch3x3dbl = self.brach3x3dbl_2(branch3x3dbl)
        branch3x3dbl = self.brach3x3dbl_3(branch3x3dbl)

        branch_pool = F.max_pool2d(x, kernel_size=3, stride=2)
        outputs = [branch3x3, branch3x3dbl, branch_pool]
        return torch.cat(outputs, 1)

class InceptionC(nn.Module):

    def __init__(self, in_channels, channel_7x7):
        super(InceptionC, self).__init__()
        self.branch1x1 = BasicConv2d(in_channels, 192, kernel_size=1)

        c7 = channel_7x7
        self.branch7x7_1 = BasicConv2d(in_channels, c7, kernel_size=1)
        self.branch7x7_2 = BasicConv2d(c7, c7, kernel_size=(1, 7), padding=(0, 3))
        self.branch7x7_3 = BasicConv2d(c7, 192, kernel_size=(7, 1), padding=(3, 0))

        self.branch7x7dbl_1 = BasicConv2d(in_channels, c7, kernel_size=1)
        self.branch7x7dbl_2 = BasicConv2d(c7, c7, kernel_size=(7, 1), padding=(3, 0))
        self.branch7x7dbl_3 = BasicConv2d(c7, c7, kernel_size=(1, 7), padding=(0, 3))
        self.branch7x7dbl_4 = BasicConv2d(c7, c7, kernel_size=(7, 1), padding=(3, 0))
        self.branch7x7dbl_5 = BasicConv2d(c7, 192, kernel_size=(1, 7), padding=(0, 3))

        self.branch_pool = BasicConv2d(in_channels, 192, kernel_size=1)

    def forward(self, x):
        branch1x1 = self.branch1x1(x)

        branch7x7 = self.branch7x7_1(x)
        branch7x7 = self.branch7x7_2(branch7x7)
        branch7x7 = self.branch7x7_3(branch7x7)

        branch7x7dbl = self.branch7x7dbl_1(x)
        branch7x7dbl = self.branch7x7dbl_2(branch7x7dbl)
        branch7x7dbl = self.branch7x7dbl_3(branch7x7dbl)
        branch7x7dbl = self.branch7x7dbl_4(branch7x7dbl)
        branch7x7dbl = self.branch7x7dbl_5(branch7x7dbl)

        branch_pool = F.avg_pool2d(x, kernel_size=3, stride=1, padding=1)
        branch_pool = self.branch_pool(branch_pool)

        outputs = [branch1x1, branch7x7, branch7x7dbl, branch_pool]
        return torch.cat(outputs, 1)

class InceptionD(nn.Module):

    def __init__(self, in_channels):
        super(InceptionD, self).__init__()
        self.branch3x3_1 = BasicConv2d(in_channels, 192, kernel_size=1)
        self.branch3x3_2 = BasicConv2d(192, 320, kernel_size=3, stride=2)

        self.branch7x7x3_1 = BasicConv2d(in_channels, 192, kernel_size=1)
        self.branch7x7x7_2 = BasicConv2d(192, 192, kernel_size=(1, 7), padding=(0,3))
        self.branch7x7x7_3 = BasicConv2d(192, 192, kernel_size=(7, 1), padding=(3,0))
        self.branch7x7x7_4 = BasicConv2d(192, 192, kernel_size=3, stride=2)

    def forward(self, x):
        branch3x3 = self.branch3x3_1(x)
        branch3x3 = self.branch3x3_2(branch3x3)

        branch7x7x3 = self.branch7x7x3_1(x)
        branch7x7x3 = self.branch7x7x3_2(branch7x7x3)
        branch7x7x3 = self.branch7x7x3_3(branch7x7x3)
        branch7x7x3 = self.branch7x7x3_4(branch7x7x3)

        branch_pool = F.max_pool2d(x, kernel_size=3, stride=2)
        outputs = [branch3x3, branch7x7x3, branch_pool]
        return torch.cat(outputs, 1)

class InceptionE(nn.Module):

    def __init__(self, in_channels):
        super(InceptionE, self).__init__()
        self.branch1x1 = BasicConv2d(in_channels, 320, kernel_size=1)

        self.branch3x3_1 = BasicConv2d(in_channels, 384, kernel_size=1)
        self.branch3x3_2a = BasicConv2d(384, 384, kernel_size=(1, 3), padding=(0, 1))
        self.branch3x3_2b = BasicConv2d(384, 384, kernel_size=(3, 1), padding=(1, 0))

        self.branch3x3dbl_1 = BasicConv2d(in_channels, 448, kerenl_size=1)
        self.branch3x3dbl_2 = BasicConv2d(448, 384, kernel_size=3, padding=1)
        self.branch3x3dbl_3a = BasicConv2d(384, 384, kernel_size=(1, 3), padding=(0, 1))
        self.branch3x3dbl_3b = BasicConv2d(384, 384, kernel_size=(3, 1), padding=(1, 0))

        self.branch_pool = BasicConv2d(in_channels, 192, kerenl_size=1)

    def forward(self, x):
        branch1x1 = self.branch1x1(x)

        branch3x3 = self.branch3x3_1(x)
        branch_3x3 = [
            self.branch3x3_2a(branch3x3),
            self.branch3x3_2b(branch3x3)
        ]
        branch3x3 = torch.cat(branch3x3, 1)

        branch3x3dbl = self.branch3x3dbl_1(x)
        branch3x3dbl = self.branch3x3dbl_2(branch3x3dbl)
        branch3x3dbl = [
            self.branch3x3dbl_3a(branch3x3dbl),
            self.branch3x3dbl_3b(branch3x3dbl),
        ]
        branch3x3dbl = torch.cat(branch3x3dbl, 1)

        branch_pool = F.avg_pool2d(x, kernel_size=3, stride=1, padding=1)
        branch_pool = self.branch_pool(branch_pool)

        outputs = [branch1x1, branch3x3, branch3x3dbl, branch_pool]
        return torch.cat(outputs, 1)

class InceptionAux(nn.Module):

    def __init__(self, in_channels, num_classes):
        super(InceptionAux, self).__init__()
        self.conv0 = BasicConv2d(in_channels, 128, kernel_size=1)
        self.conv1 = BasicConv2d(128, 768, kernel_size=5)
        self.conv1.stddev = 0.01
        self.fc = nn.Linear(768, num_classes)
        self.fc.stddev = 0.001

    def forward(self, x):
        # N x 768 x 17 x 17
        x = F.avg_pool2d(x, kernel_size=5, stride=3)
        # N x 768 x 5 x 5
        x = self.conv0(x)
        # N x 128 x 5 x 5
        x = self.conv1(x)
        # N x 768 x 1 x 1
        x = F.adaptive_avg_pool2d(x, (1, 1))
        # N X 768 X 1 X 1
        X = x.view(x.size(0), -1)
        # N x 768
        x = self.fc(x)
        # N x 1000
        return x

class Inception3(nn.Module):

    def __init__(self, num_classes=1000, aux_logits=True, transform_input=False):
        super(Inception3, self).__init__()
        self.aux_logits = aux_logits
        self.transform_input = transform_input
        self.Conv2d_1a_3x3 = BasicConv2d(3, 32, kernel_size=3, stride=2)
        self.Conv2d_2a_3x3 = BasicConv2d(32, 32, kernel_size=3)
        self.Conv2d_2b_3x3 = BasicConv2d(32, 64, kernel_size=3, padding=1)
        self.Conv2d_3b_1x1 = BasicConv2d(64, 80, kernel_size=1)
        self.Conv2d_4a_3x3 = BasicConv2d(80, 192, kernel_size=3)
        self.Mixed_5b = InceptionA(192, pool_features=32)
        self.Mixed_5c = InceptionA(256, pool_features=64)
        self.Mixed_5d = InceptionA(288, pool_features=64)
        self.Mixed_6a = InceptionB(288)
        self.Mixed_6b = InceptionC(768, channel_7x7=128)
        self.Mixed_6c = InceptionC(768, channel_7x7=160)
        self.Mixed_6d = InceptionC(768, channel_7x7=160)
        self.Mixed_6e = InceptionC(768, channel_7x7=192)
        if aux_logits:
            self.AuxLogits = InceptionAux(768, num_classes)
        self.Mixed_7a = InceptionD(768)
        self.Mixed_7b = InceptionE(1280)
        self.Mixed_7c = InceptionE(2048)
        self.fc = nn.Linear(2048, num_classes)

        for m in self.modules():
            if isinstance(m, nn.Conv2d) or isinstance(m, nn.Linear):
                import scipy.stats as stats
                stddev = m.stddev if hasattr(m, 'stddev') else 0.1
                X = stats.truncnorm(-2, 2, scale=stddev)
                values = torch.as_tensor(X.rvs(m.weight.numel()), dtype=m.weight.dtype)
                values = values.view(m.weight.size())
                with torch.no_grad():
                    m.weight.copy_(values)
            elif isinstance(m, nn.BatchNorm2d):
                nn.init.constant_(m.weight, 1)
                nn.init.constant_(m.bias, 0)

    def forward(self, x):
        if self.transform_input:
            x_ch0 = torch.unsqueeze(x[:, 0], 1) * (0.229 / 0.5) + (0.485 - 0.5) / 0.5
            x_ch1 = torch.unsqueeze(x[:, 1], 1) * (0.224 / 0.5) + (0.456 - 0.5) / 0.5
            x_ch2 = torch.unsqueeze(x[:, 2], 1) * (0.225 / 0.5) + (0.406 - 0.5) / 0.5
            x = torch.cat((x_ch0, x_ch1, x_ch2), 1)
        # N x 3 x 299 x 299
        x = self.Conv2d_1a_3x3(x)
        # N x 32 x 149 x 149
        x = self.Conv2d_2a_3x3(x)
        # N x 32 x 147 x 147
        x = self.Conv2d_2b_3x3(x)
        # N x 64 x 147 x 147
        x = F.max_pool2d(x, kernel_size=3, stride=2)
        # N x 64 x 73 x 73
        x = self.Conv2d_3b_1x1(x)
        # N x 80 x 73 x 73
        x = self.Conv2d_4a_3x3(x)
        # N x 192 x 71 x 71
        x = F.max_pool2d(x, kernel_size=3, stride=2)
        # N x 192 x 35 x 35
        x = self.Mixed_5b(x) # 32+64+64+96=256
        # N x 256 x 35 x 35
        x = self.Mixed_5c(x) # 64+64+64+96=288
        # N x 288 x 35 x 35
        x = self.Mixed_5d(x) # 64+64+64+96=288
        # N x 288 x 35 x 35
        x = self.Mixed_6a(x) # 384+96+288=768
        # N x 768 x 17 x 17
        x = self.Mixed_6b(x) # 192+192+192+192=768
        # N x 768 x 17 x 17
        x = self.Mixed_6c(x) # 192+192+192+192=768
        # N x 768 x 17 x 17
        x = self.Mixed_6d(x) # 192+192+192+192=768
        # N x 768 x 17 x 17
        x = self.Mixed_6e(x) # 192+192+192+192=768
        # N x 768 x 17 x 17
        if self.training and self.aux_logits:
            aux = self.AuxLogits(x) #fc->768
        # N x 768 x 17 x 17
        x = self.Mixed_7a(x) # 320+192+768=1280
        # N x 1280 x 8 x 8
        x = self.Mixed_7b(x) #320+384x2+384x2+192=2048
        # N x 2048 x 8 x 8
        x = self.Mixed_7c(x) #320+384x2+384x2+192=2048
        # N x 2048 x 8 x 8
        # Adaptive average pooling
        x = F.adaptive_avg_pool2d(x, (1,1))
        # N x 2048 x 1 x 1
        x = F.dropout(x, training=self.training)
        # N x 2048 x 1 x 1
        x = x.view(x.size(0), -1)
        # N x 2048
        x = self.fc(x)
        # N x 1000
        if self.training and self.aux_logits:
            return _InceptionOuputs(x, aux)
        return x

def inception_v3(pretrained=False, progress=True, **kwargs):
    if pretrained:
        if 'transform_input' not in kwargs:
            kwargs['transform_input'] = True
        if 'aux_logits' in kwargs:
            original_aux_logits = kwargs['aux_logits']
            kwargs['aux_logits'] = True
        else:
            original_aux_logits = True
        model = Inception3(**kwargs)
        state_dict = load_state_dict_from_url(model_urls['inception_v3_google'],
                                              progress=progress)
        model.load_state_dict(state_dict)
        if not original_aux_logits:
            model.aux_logits = False
            del model.AuxLogits
        return model
    return Inception3(**kwargs)

二、一些有趣的用法

2.1 torch.nn.ReLU和torch.nn.functional.relu区别

 实际计算结果一致。不同的是nn.ReLU作为一个层结构,必须添加到nn.Module容器中才能使用,而F.ReLU则作为一个函数调用。

F.relu(x, inplace=True)
x = F.max_pool2d(x, kernel_size=3, stride=2)
x = F.adaptive_avg_pool2d(x, (1,1))
  

2.2 设置BN的参数

 self.bn = nn.BatchNorm2d(out_channels, eps=0.001)

2.3 按channel拼接的一种方式

	branch1x1 = ...
	branch5x5 = ...
	branch3x3dbl = ...
	branchpool = ...
	outputs = [branch1x1, branch5x5, branch3x3dbl, branch_pool]
    return torch.cat(outputs, 1)

注意,torch维度默认Batch_size x Channel x W x H,因此选择在第1维度拼接.

2.4 不对称卷积操作

即kernel_size不再是正方形.
首先是两个不对称卷积的串联，即首先nx1卷积，再1xn卷积.

        self.branch7x7dbl_1 = BasicConv2d(in_channels, c7, kernel_size=1)
        self.branch7x7dbl_2 = BasicConv2d(c7, c7, kernel_size=(7, 1), padding=(3, 0))
        self.branch7x7dbl_3 = BasicConv2d(c7, c7, kernel_size=(1, 7), padding=(0, 3))
        self.branch7x7dbl_4 = BasicConv2d(c7, c7, kernel_size=(7, 1), padding=(3, 0))
        self.branch7x7dbl_5 = BasicConv2d(c7, 192, kernel_size=(1, 7), padding=(0, 3))

然后是两个不对称卷积的并联，即分别对输入feature进行nx1卷积和1xn卷积,再进行concat操作.

def __init__(self, in_channels):
        ...
        self.branch3x3dbl_1 = BasicConv2d(in_channels, 448, kerenl_size=1)
       self.branch3x3dbl_2 = BasicConv2d(448, 384, kernel_size=3, padding=1)
       self.branch3x3dbl_3a = BasicConv2d(384, 384, kernel_size=(1, 3), padding=(0, 1))
       self.branch3x3dbl_3b = BasicConv2d(384, 384, kernel_size=(3, 1), padding=(1, 0))
      	...
def forward(self, x):
		...
		 branch3x3dbl = self.branch3x3dbl_1(x)
       branch3x3dbl = self.branch3x3dbl_2(branch3x3dbl)
       branch3x3dbl = [
           self.branch3x3dbl_3a(branch3x3dbl),
           self.branch3x3dbl_3b(branch3x3dbl),
       ]
       branch3x3dbl = torch.cat(branch3x3dbl, 1)
       ...

2.5 中间层输出作为辅助输出

_InceptionOuputs = namedtuple('InceptionOuputs', ['logits', 'aux_logits'])
class Inception(..):
   ..
 	def __init__(self, x):
 		...
 		x = self.Mixed_6e(x) # 192+192+192+192=768
       # N x 768 x 17 x 17
       if self.training and self.aux_logits:
           aux = self.AuxLogits(x) #fc->768
       x = ...
        if self.training and self.aux_logits:
           return _InceptionOuputs(x, aux)
       return x          

namedtuple能够用来创建类似于元组的数据类型，除了能够用索引来访问数据，能够迭代，更能够方便的通过属性名来访问数据。

2.6 在前向传播中对输入图像进行预训练的归一化

        if self.transform_input: #torch.y=unsqueeze()用来将(batch_size, 1, w, h)->(batch_size, w, h)
            x_ch0 = torch.unsqueeze(x[:, 0], 1) * (0.229 / 0.5) + (0.485 - 0.5) / 0.5 #R
            x_ch1 = torch.unsqueeze(x[:, 1], 1) * (0.224 / 0.5) + (0.456 - 0.5) / 0.5 #G
            x_ch2 = torch.unsqueeze(x[:, 2], 1) * (0.225 / 0.5) + (0.406 - 0.5) / 0.5 #B
            x = torch.cat((x_ch0, x_ch1, x_ch2), 1) #在RGB通道进行concat操作

2.7 一种flatten的方法

        # N x 2048 x 1 x 1
        x = x.view(x.size(0), -1) #即将x resize为(batch_size, -1),-1为自动计算剩余维度
        # N x 2048