Baselines/Models/HiCSR.py

# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.
# --------------------------------------------------------
# References:
# HiCSR: https://github.com/PSI-Lab/HiCSR
# --------------------------------------------------------
import torch
from torch import nn
import torch.nn.functional as F
import math


class ResidualBlock(nn.Module):
    def __init__(self, channels):
        super(ResidualBlock, self).__init__()
        self.conv1 = nn.Conv2d(channels, channels, kernel_size=3, padding=1)
        self.bn1 = nn.BatchNorm2d(channels)
        self.relu = nn.ReLU()
        self.conv2 = nn.Conv2d(channels, channels, kernel_size=3, padding=1)
        self.bn2 = nn.BatchNorm2d(channels)

    def forward(self, x):
        res = self.conv1(x)
        res = self.bn1(res)
        res = self.relu(res)
        res = self.conv2(res)
        res = self.bn2(res)
        return x + res


class Generator(nn.Module):
    def __init__(self, num_res_blocks=15, input_channels=1, out_channels=None):
        super(Generator, self).__init__()
        if not out_channels:
            out_channels = input_channels
        self.pre_res_block = nn.Sequential(
                nn.Conv2d(input_channels, 64, kernel_size=3),
                nn.ReLU(),
                )
                
        res_blocks = [ResidualBlock(64) for _ in range(num_res_blocks)]
        self.res_blocks = nn.Sequential(*res_blocks)

        self.post_res_block = nn.Sequential(
                nn.Conv2d(64, 64, kernel_size=3, padding=1),
                nn.BatchNorm2d(64)
                )

        self.final_block = nn.Sequential(
                nn.Conv2d(64, 128, kernel_size=3),
                nn.Conv2d(128, 128, kernel_size=3),
                nn.Conv2d(128, 256, kernel_size=3),
                nn.Conv2d(256, 256, kernel_size=3),
                nn.Conv2d(256, out_channels, kernel_size=3),
                )

    def forward(self, x):
        first_block = self.pre_res_block(x)
        res_blocks = self.res_blocks(first_block)
        post_res_block = self.post_res_block(res_blocks)
        final_block = self.final_block(first_block + post_res_block)
        return torch.tanh(final_block)


class Discriminator(nn.Module):
    def __init__(self, input_channels=1):
        super(Discriminator, self).__init__()
        self.conv = nn.Sequential(
                nn.Conv2d(input_channels, 64, kernel_size=4, stride=2, padding=1, bias=False),
                nn.LeakyReLU(0.2, inplace=True),

                nn.Conv2d(64, 128, kernel_size=4, stride=2, padding=1, bias=False),
                nn.BatchNorm2d(128),
                nn.LeakyReLU(0.2, inplace=True),

                nn.Conv2d(128, 256, kernel_size=4, stride=2, padding=1, bias=False),
                nn.BatchNorm2d(256),
                nn.LeakyReLU(0.2, inplace=True),

                nn.Conv2d(256, 512, kernel_size=4, stride=2, padding=1, bias=False),
                nn.BatchNorm2d(512),
                nn.LeakyReLU(0.2, inplace=True),

                nn.Conv2d(512, input_channels, kernel_size=1, stride=1, padding=0, bias=False),
                nn.Sigmoid())# temporarily add the sigmoid
        self.init_params()
        
    def forward(self, x):
        x = self.conv(x)
        return x
    
    def init_params(self):
        for module in self.modules():
            if isinstance(module, nn.Conv2d):
                nn.init.normal_(module.weight.data, 0.0, 0.02)
            elif isinstance(module, nn.BatchNorm2d):
                nn.init.normal_(module.weight.data, 1.0, 0.02)
                nn.init.constant_(module.bias.data, 0)

class DAE(nn.Module):
    def __init__(self, num_layers=5, num_features=64, input_channels=1):
        super(DAE, self).__init__()
        self.num_layers = num_layers

        conv_layers = []
        deconv_layers = []

        conv_layers.append(nn.Sequential(nn.Conv2d(input_channels, num_features, kernel_size=3, stride=2, padding=1),
                                         nn.ReLU(inplace=True)))
        for i in range(num_layers - 1):
            conv_layers.append(nn.Sequential(nn.Conv2d(num_features, num_features, kernel_size=3, padding=1),
                                             nn.ReLU(inplace=True)))

        for i in range(num_layers - 1):
            deconv_layers.append(nn.Sequential(nn.ConvTranspose2d(num_features, num_features, kernel_size=3, padding=1),
                                               nn.ReLU(inplace=True)))
        deconv_layers.append(nn.ConvTranspose2d(num_features, input_channels, kernel_size=3, stride=2, padding=1, output_padding=1))

        self.conv_layers = nn.Sequential(*conv_layers)
        self.deconv_layers = nn.Sequential(*deconv_layers)
        self.relu = nn.ReLU(inplace=True)

    def forward(self, x):
        residual = x

        conv_feats = []
        for i in range(self.num_layers):
            x = self.conv_layers[i](x)
            if (i + 1) % 2 == 0 and len(conv_feats) < math.ceil(self.num_layers / 2) - 1:
                conv_feats.append(x)

        conv_feats_idx = 0
        for i in range(self.num_layers):
            x = self.deconv_layers[i](x)
            if (i + 1 + self.num_layers) % 2 == 0 and conv_feats_idx < len(conv_feats):
                conv_feat = conv_feats[-(conv_feats_idx + 1)]
                conv_feats_idx += 1
                x = x + conv_feat
                x = self.relu(x)

        x += residual
        x = torch.tanh(x)

        return x