model.py

import torch
import torch.nn as nn


class ConvBlock(nn.Module):

    def __init__(self, in_channels, out_channels):
        super().__init__()

        self.block = nn.Sequential(
            nn.Conv2d(
                in_channels,
                out_channels,
                kernel_size=3,
                padding=1
            ),
            nn.ReLU(inplace=True),

            nn.Conv2d(
                out_channels,
                out_channels,
                kernel_size=3,
                padding=1
            ),
            nn.ReLU(inplace=True)
        )

    def forward(self, x):
        return self.block(x)


class Encoder(nn.Module):

    def __init__(self):
        super().__init__()

        self.stage_1 = ConvBlock(
            in_channels=3,
            out_channels=9
        )

        self.stage_2 = ConvBlock(
            in_channels=9,
            out_channels=27
        )

        self.stage_3 = nn.Conv2d(
            in_channels=27,
            out_channels=3,
            kernel_size=3,
            padding=1
        )

    def forward(self, x):

        x = self.stage_1(x)
        x = self.stage_2(x)
        x = self.stage_3(x)

        return x


class Decoder(nn.Module):

    def __init__(self):
        super().__init__()

        self.upsample_block = nn.Upsample(
            size=(640, 1024),
            mode="bilinear",
            align_corners=False
        )

    def forward(self, x):
        return self.upsample_block(x)


class ImageRegressionNet(nn.Module):

    def __init__(self):
        super().__init__()

        self.encoder = Encoder()

        self.decoder = Decoder()

        self.encoder_decoder = nn.Sequential(
            self.encoder,
            self.decoder
        )

    def forward(self, x):
        return self.encoder_decoder(x)