pytorch在fintune时将sequential中的层输出，以vgg为例

pytorch将sequential中的层输出，以vgg为例

有时候我们在fintune时发现pytorch把许多层都集合在一个sequential里，但是我们希望能把中间层的结果引出来做下一步操作，于是我自己琢磨了一个方法，以vgg为例，有点僵硬哈！

首先pytorch自带的vgg16模型的网络结构如下：

VGG(
  (features): Sequential(
    (0): Conv2d (3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (1): ReLU(inplace)
    (2): Conv2d (64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (3): ReLU(inplace)
    (4): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1))
    (5): Conv2d (64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (6): ReLU(inplace)
    (7): Conv2d (128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (8): ReLU(inplace)
    (9): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1))
    (10): Conv2d (128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (11): ReLU(inplace)
    (12): Conv2d (256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (13): ReLU(inplace)
    (14): Conv2d (256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (15): ReLU(inplace)
    (16): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1))
    (17): Conv2d (256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (18): ReLU(inplace)
    (19): Conv2d (512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (20): ReLU(inplace)
    (21): Conv2d (512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (22): ReLU(inplace)
    (23): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1))
    (24): Conv2d (512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (25): ReLU(inplace)
    (26): Conv2d (512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (27): ReLU(inplace)
    (28): Conv2d (512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (29): ReLU(inplace)
    (30): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1))
  )
  (classifier): Sequential(
    (0): Linear(in_features=25088, out_features=4096)
    (1): ReLU(inplace)
    (2): Dropout(p=0.5)
    (3): Linear(in_features=4096, out_features=4096)
    (4): ReLU(inplace)
    (5): Dropout(p=0.5)
    (6): Linear(in_features=4096, out_features=1000)
  )
)

我们需要fintune vgg16的features部分，并且我希望把3，8， 15， 22， 29这五个作为输出进一步操作。我的想法是自己写一个vgg网络，这个网络参数与pytorch的网络一致但是保证我们需要的层输出在sequential外。于是我写的网络如下：

class our_vgg(nn.Module):
    def __init__(self):
        super(our_vgg, self).__init__()
        self.conv1 = nn.Sequential(
            # conv1
            nn.Conv2d(3, 64, 3, padding=35),
            nn.ReLU(inplace=True),
            nn.Conv2d(64, 64, 3, padding=1),
            nn.ReLU(inplace=True),

        )
        self.conv2 = nn.Sequential(
            # conv2
            nn.MaxPool2d(2, stride=2, ceil_mode=True),  # 1/2
            nn.Conv2d(64, 128, 3, padding=1),
            nn.ReLU(inplace=True),
            nn.Conv2d(128, 128, 3, padding=1),
            nn.ReLU(inplace=True),

        )
        self.conv3 = nn.Sequential(
            # conv3
            nn.MaxPool2d(2, stride=2, ceil_mode=True),  # 1/4
            nn.Conv2d(128, 256, 3, padding=1),
            nn.ReLU(inplace=True),
            nn.Conv2d(256, 256, 3, padding=1),
            nn.ReLU(inplace=True),
            nn.Conv2d(256, 256, 3, padding=1),
            nn.ReLU(inplace=True),

        )
        self.conv4 = nn.Sequential(
            # conv4
            nn.MaxPool2d(2, stride=2, ceil_mode=True),  # 1/8
            nn.Conv2d(256, 512, 3, padding=1),
            nn.ReLU(inplace=True),
            nn.Conv2d(512, 512, 3, padding=1),
            nn.ReLU(inplace=True),
            nn.Conv2d(512, 512, 3, padding=1),
            nn.ReLU(inplace=True),

        )
        self.conv5 = nn.Sequential(
            # conv5
            nn.MaxPool2d(2, stride=2, ceil_mode=True),  # 1/16
            nn.Conv2d(512, 512, 3, padding=1),
            nn.ReLU(inplace=True),
            nn.Conv2d(512, 512, 3, padding=1),
            nn.ReLU(inplace=True),
            nn.Conv2d(512, 512, 3, padding=1),
            nn.ReLU(inplace=True),
        )


    def forward(self, x):

        conv1 = self.conv1(x)
        conv2 = self.conv2(conv1)
        conv3 = self.conv3(conv2)
        conv4 = self.conv4(conv3)
        conv5 = self.conv5(conv4)

        return conv5

接着就是copy weights了：

def convert_vgg(vgg16):#vgg16是pytorch自带的
    net = our_vgg()# 我写的vgg

    vgg_items = net.state_dict().items()
    vgg16_items = vgg16.items()

    pretrain_model = {}
    j = 0
    for k, v in net.state_dict().iteritems():#按顺序依次填入
        v = vgg16_items[j][1]
        k = vgg_items[j][0]
        pretrain_model[k] = v
        j += 1
    return pretrain_model


## net是我们最后使用的网络，也是我们想要放置weights的网络
net = net()

print ('load the weight from vgg')
pretrained_dict = torch.load('vgg16.pth')
pretrained_dict = convert_vgg(pretrained_dict)
model_dict = net.state_dict()
# 1. 把不属于我们需要的层剔除
pretrained_dict = {k: v for k, v in pretrained_dict.items() if k in model_dict}
# 2. 把参数存入已经存在的model_dict
model_dict.update(pretrained_dict) 
# 3. 加载更新后的model_dict
net.load_state_dict(model_dict)
print ('copy the weight sucessfully')

这样我就基本达成目标了，注意net也就是我们要使用的网络fintune部分需要和our_vgg一致。