efficientnet.py

"""
MIT License

Copyright (c) 2017 liukuang

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

EfficientNet in PyTorch.

Paper: "EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks".
"""
import torch
import torch.nn as nn
import torch.nn.functional as F


class Block(nn.Module):
    '''expand + depthwise + pointwise + squeeze-excitation'''

    def __init__(self, in_planes, out_planes, expansion, stride):
        super(Block, self).__init__()
        self.stride = stride

        planes = expansion * in_planes
        self.conv1 = nn.Conv2d(
            in_planes, planes, kernel_size=1, stride=1, padding=0, bias=False)
        self.bn1 = nn.BatchNorm2d(planes)
        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3,
                               stride=stride, padding=1, groups=planes, bias=False)
        self.bn2 = nn.BatchNorm2d(planes)
        self.conv3 = nn.Conv2d(
            planes, out_planes, kernel_size=1, stride=1, padding=0, bias=False)
        self.bn3 = nn.BatchNorm2d(out_planes)

        self.shortcut = nn.Sequential()
        if stride == 1 and in_planes != out_planes:
            self.shortcut = nn.Sequential(
                nn.Conv2d(in_planes, out_planes, kernel_size=1,
                          stride=1, padding=0, bias=False),
                nn.BatchNorm2d(out_planes),
            )

        # SE layers
        self.fc1 = nn.Conv2d(out_planes, out_planes//16, kernel_size=1)
        self.fc2 = nn.Conv2d(out_planes//16, out_planes, kernel_size=1)

    def forward(self, x):
        out = F.relu(self.bn1(self.conv1(x)))
        out = F.relu(self.bn2(self.conv2(out)))
        out = self.bn3(self.conv3(out))
        shortcut = self.shortcut(x) if self.stride == 1 else out
        # Squeeze-Excitation
        w = F.avg_pool2d(out, out.size(2))
        w = F.relu(self.fc1(w))
        w = self.fc2(w).sigmoid()
        out = out * w + shortcut
        return out


class EfficientNet(nn.Module):
    def __init__(self, cfg, num_classes=10):
        super(EfficientNet, self).__init__()
        self.cfg = cfg
        self.conv1 = nn.Conv2d(3, 32, kernel_size=3,
                               stride=1, padding=1, bias=False)
        self.bn1 = nn.BatchNorm2d(32)
        self.layers = self._make_layers(in_planes=32)
        self.linear = nn.Linear(cfg[-1][1], num_classes)

    def _make_layers(self, in_planes):
        layers = []
        for expansion, out_planes, num_blocks, stride in self.cfg:
            strides = [stride] + [1]*(num_blocks-1)
            for stride in strides:
                layers.append(Block(in_planes, out_planes, expansion, stride))
                in_planes = out_planes
        return nn.Sequential(*layers)

    def forward(self, x):
        out = F.relu(self.bn1(self.conv1(x)))
        out = self.layers(out)
        out = out.view(out.size(0), -1)
        out = self.linear(out)
        return out


def EfficientNetB0(num_classes: int = 10):
    # (expansion, out_planes, num_blocks, stride)
    cfg = [(1,  16, 1, 2),
           (6,  24, 2, 1),
           (6,  40, 2, 2),
           (6,  80, 3, 2),
           (6, 112, 3, 1),
           (6, 192, 4, 2),
           (6, 320, 1, 2)]
    return EfficientNet(cfg, num_classes=num_classes)


def test():
    net = EfficientNetB0()
    x = torch.randn(2, 3, 32, 32)
    y = net(x)
    print(y.shape)


# test()