You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
22 lines
716 B
22 lines
716 B
5 years ago
|
from torch import nn as nn
|
||
|
|
|
||
|
|
|
||
|
|
class SEModule(nn.Module):
|
||
|
|
|
||
|
|
def __init__(self, channels, reduction=16, act_layer=nn.ReLU):
|
||
|
|
super(SEModule, self).__init__()
|
||
|
|
self.avg_pool = nn.AdaptiveAvgPool2d(1)
|
||
|
|
reduction_channels = max(channels // reduction, 8)
|
||
|
|
self.fc1 = nn.Conv2d(
|
||
|
|
channels, reduction_channels, kernel_size=1, padding=0, bias=True)
|
||
|
|
self.act = act_layer(inplace=True)
|
||
|
|
self.fc2 = nn.Conv2d(
|
||
|
|
reduction_channels, channels, kernel_size=1, padding=0, bias=True)
|
||
|
|
|
||
|
|
def forward(self, x):
|
||
|
|
x_se = self.avg_pool(x)
|
||
|
|
x_se = self.fc1(x_se)
|
||
|
|
x_se = self.act(x_se)
|
||
|
|
x_se = self.fc2(x_se)
|
||
|
|
return x * x_se.sigmoid()
|