Public_Classifier.py

# based on vgg16

import torch
import torch.nn as nn


class Public_Classifier(nn.Module):

    def __init__(self, output=1):
        super(Public_Classifier, self).__init__()

        self.features3 = nn.Sequential(

            nn.Conv2d(128, 256, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True),
            nn.ReLU(inplace=True),
            nn.Conv2d(256, 256, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True),
            nn.ReLU(inplace=True),
            nn.Conv2d(256, 256, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True),
            nn.ReLU(inplace=True)
        )

        self.pool3 = nn.MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False, return_indices=True)

        self.features4 = nn.Sequential(

            nn.Conv2d(256, 512, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True),
            nn.ReLU(inplace=True),
            nn.Conv2d(512, 512, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True),
            nn.ReLU(inplace=True),
            nn.Conv2d(512, 512, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True),
            nn.ReLU(inplace=True)
        )

        self.pool4 = nn.MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False, return_indices=True)

        self.features5 = nn.Sequential(

            nn.Conv2d(512, 512, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True),
            nn.ReLU(inplace=True),
            nn.Conv2d(512, 512, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True),
            nn.ReLU(inplace=True),
            nn.Conv2d(512, 512, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True),
            nn.ReLU(inplace=True)
        )

        self.pool5 = nn.MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False, return_indices=True)

        self.classifier = nn.Sequential(
            nn.Linear(15360, 4096, bias=True),
            nn.ReLU(inplace=True),
            nn.Dropout(p=0.5),
            nn.Linear(4096, 4096, bias=True),
            nn.ReLU(inplace=True),
            nn.Dropout(p=0.5),
            nn.Linear(4096, output, bias=True),
            nn.Sigmoid()
        )

    def forward(self, x):

        p3 = self.features3(x)
        x, p3_idx = self.pool3(p3)
        p4 = self.features4(x)
        x, p4_idx = self.pool4(p4)
        p5 = self.features5(x)
        x, p5_idx = self.pool5(p5)
        x = x.view(x.size(0), -1)
        out = self.classifier(x)

        return out