Resnet Variational autoencoder for image reconstruction

vae_model.py

import torch
from torch import nn
import torch.nn.functional as F
import abc
import pytorch_ssim
import torchvision.models as models
from torch.autograd import Variable


class AbstractAutoEncoder(nn.Module):
    __metaclass__ = abc.ABCMeta

    @abc.abstractmethod
    def encode(self, x):
        return

    @abc.abstractmethod
    def decode(self, z):
        return

    @abc.abstractmethod
    def forward(self, x, latent_vec=False):
        """model return (reconstructed_x, *)"""
        return

    @abc.abstractmethod
    def loss_function(self, **kwargs):
        """accepts (original images, *) where * is the same as returned from forward()"""
        return

    @abc.abstractmethod
    def latest_losses(self):
        """returns the latest losses in a dictionary. Useful for logging."""
        return

class ResNet_VAE(AbstractAutoEncoder):
    def __init__(
        self,recon_loss_type, fc_hidden1=1024,
            fc_hidden2=768, drop_p=0.3, CNN_embed_dim=256):
        super(ResNet_VAE, self).__init__()
        self.recon_loss_type = recon_loss_type
        self.fc_hidden1, self.fc_hidden2, self.CNN_embed_dim = fc_hidden1, fc_hidden2, CNN_embed_dim

        # CNN architechtures
        self.ch1, self.ch2, self.ch3, self.ch4 = 16, 32, 64, 128
        self.k1, self.k2, self.k3, self.k4 = (5, 5), (3, 3), (3, 3), (3, 3)      # 2d kernal size
        self.s1, self.s2, self.s3, self.s4 = (2, 2), (2, 2), (2, 2), (2, 2)      # 2d strides
        self.pd1, self.pd2, self.pd3, self.pd4 = (0, 0), (0, 0), (0, 0), (0, 0)  # 2d padding

        # encoding components
        resnet = models.resnet18(pretrained=True)
        modules = list(resnet.children())[:-1]      # delete the last fc layer.
        self.resnet = nn.Sequential(*modules)
        self.fc1 = nn.Linear(resnet.fc.in_features, self.fc_hidden1)
        self.bn1 = nn.BatchNorm1d(self.fc_hidden1, momentum=0.01)
        self.fc2 = nn.Linear(self.fc_hidden1, self.fc_hidden2)
        self.bn2 = nn.BatchNorm1d(self.fc_hidden2, momentum=0.01)
        # Latent vectors mu and sigma
        self.fc3_mu = nn.Linear(self.fc_hidden2, self.CNN_embed_dim)      # output = CNN embedding latent variables
        self.fc3_logvar = nn.Linear(self.fc_hidden2, self.CNN_embed_dim)  # output = CNN embedding latent variables

        # Sampling vector
        self.fc4 = nn.Linear(self.CNN_embed_dim, self.fc_hidden2)
        self.fc_bn4 = nn.BatchNorm1d(self.fc_hidden2)
        self.fc5 = nn.Linear(self.fc_hidden2, 64 * 4 * 4)
        self.fc_bn5 = nn.BatchNorm1d(64 * 4 * 4)
        self.relu = nn.ReLU(inplace=True)

        # Decoder
        self.convTrans9 = nn.Sequential(
            nn.ConvTranspose2d(in_channels=1024, out_channels=512, kernel_size=self.k4, stride=self.s4,
                               padding=self.pd4),
            nn.BatchNorm2d(512, momentum=0.01),
            nn.ReLU(inplace=True),
        )
        self.convTrans10 = nn.Sequential(
            nn.ConvTranspose2d(in_channels=512, out_channels=256, kernel_size=self.k4, stride=self.s4,
                               padding=self.pd4),
            nn.BatchNorm2d(256, momentum=0.01),
            nn.ReLU(inplace=True),
        )
        self.convTrans11 = nn.Sequential(
            nn.ConvTranspose2d(in_channels=256, out_channels=128, kernel_size=self.k3, stride=self.s3,
                               padding=self.pd3),
            nn.BatchNorm2d(128, momentum=0.01),
            nn.ReLU(inplace=True),
        )
        self.convTrans12 = nn.Sequential(
            nn.ConvTranspose2d(in_channels=128, out_channels=64, kernel_size=self.k4, stride=self.s4,
                               padding=self.pd4),
            nn.BatchNorm2d(64, momentum=0.01),
            nn.ReLU(inplace=True),
        )
        self.convTrans6 = nn.Sequential(
            nn.ConvTranspose2d(in_channels=64, out_channels=32, kernel_size=self.k4, stride=self.s4,
                               padding=self.pd4),
            nn.BatchNorm2d(32, momentum=0.01),
            nn.ReLU(inplace=True),
        )
        self.convTrans7 = nn.Sequential(
            nn.ConvTranspose2d(in_channels=32, out_channels=8, kernel_size=self.k3, stride=self.s3,
                               padding=self.pd3),
            nn.BatchNorm2d(8, momentum=0.01),
            nn.ReLU(inplace=True),
        )

        self.convTrans8 = nn.Sequential(
            nn.ConvTranspose2d(in_channels=8, out_channels=3, kernel_size=self.k2, stride=self.s2,
                               padding=self.pd2),
            nn.BatchNorm2d(3, momentum=0.01),
            nn.Sigmoid()    # y = (y1, y2, y3) \in [0 ,1]^3
        )

    def encode(self, x):
        x = self.resnet(x)  # ResNet
        x = x.view(x.size(0), -1)  # flatten output of conv

        # FC layers
        x = self.bn1(self.fc1(x))
        x = self.relu(x)
        x = self.bn2(self.fc2(x))
        x = self.relu(x)
        # x = F.dropout(x, p=self.drop_p, training=self.training)
        mu, logvar = self.fc3_mu(x), self.fc3_logvar(x)
        return mu, logvar
    def reparameterize(self, mu, logvar):
        std = torch.exp(logvar/2)
        eps = torch.randn_like(std)
        return mu + eps * std
    def decode(self, z):
        x = self.fc_bn4(self.fc4(z))
        x = self.relu(x)
        x = self.fc_bn5(self.fc5(x))
        x = self.relu(x).view(-1, 1024, 1, 1)

        x = self.convTrans9(x)
        x = self.convTrans10(x)
        x = self.convTrans11(x)
        x = self.convTrans12(x)
        x = self.convTrans6(x)
        x = self.convTrans7(x)
        x = self.convTrans8(x)
        x = F.interpolate(x, size=(224, 224), mode='bilinear')
        return x
    def forward(self, x,latent_vec=False):
        mu, logvar = self.encode(x)
        z = self.reparameterize(mu, logvar)
        x_reconst = self.decode(z)
        if latent_vec:
            return x_reconst, mu, logvar, z
        else:
            return x_reconst, mu, logvar
    def loss_function(self, x, x_hat, mu, logvar):
        recon_loss = calc_reconstruction_loss(x, x_hat, self.recon_loss_type)
        kl_loss = -0.5 * torch.mean(1 + logvar - mu**2 - logvar.exp())
        return kl_loss + recon_loss
   
class ResNet_CVAE(AbstractAutoEncoder):
    def __init__(
        self,recon_loss_type, fc_hidden1=1024,
            fc_hidden2=768, drop_p=0.3, CNN_embed_dim=256):
        super(ResNet_VAE, self).__init__()
        self.recon_loss_type = recon_loss_type
        self.fc_hidden1, self.fc_hidden2, self.CNN_embed_dim = fc_hidden1, fc_hidden2, CNN_embed_dim

        # CNN architechtures
        self.ch1, self.ch2, self.ch3, self.ch4 = 16, 32, 64, 128
        self.k1, self.k2, self.k3, self.k4 = (5, 5), (3, 3), (3, 3), (3, 3)      # 2d kernal size
        self.s1, self.s2, self.s3, self.s4 = (2, 2), (2, 2), (2, 2), (2, 2)      # 2d strides
        self.pd1, self.pd2, self.pd3, self.pd4 = (0, 0), (0, 0), (0, 0), (0, 0)  # 2d padding

        # encoding components
        resnet = models.resnet18(pretrained=True)
        modules = list(resnet.children())[:-2]      # delete the last fc layer.
        self.resnet = nn.Sequential(*modules)
        self.conv1 = nn.Sequential(
            nn.Conv2d(in_channels=512, out_channels=256, kernel_size=1, stride=1,
                               padding=self.pd4),
            nn.BatchNorm2d(256, momentum=0.01),
            nn.ReLU(inplace=True),
        )
        self.out_mu = nn.Conv2d(256, 256, kernel_size=1, stride=1)
        self.out_logvar = nn.Conv2d(256, 256, kernel_size=1, stride=1)

        # (256, 8, 8) -> (3, 256, 256)
        self.decoder = nn.Sequential(
            #  (256, 8, 8) -> (256, 16, 16)
            UpsamplingLayer(256, 256, activation="ReLU"),
            # -> (128, 32, 32)
            UpsamplingLayer(256, 128, activation="ReLU"),
            # -> (64, 64, 64)
            UpsamplingLayer(128, 64, activation="ReLU", type="upsample"),
            # -> (32, 128, 128)
            UpsamplingLayer(64, 32, activation="ReLU", bn=False),
            # -> (3, 256, 256)
            UpsamplingLayer(32, 3, activation="none", bn=False, type="upsample"),
            # nn.Tanh()
            nn.Hardtanh(-1.0, 1.0),
        )
    def encode(self, x):
        x = self.resnet(x)  # ResNet
        x = self.conv1(x)
        mu = self.out_mu(x)
        logvar = self.out_logvar(x)
        return mu, logvar

    def reparameterize(self, mu, logvar):
        std = torch.exp(logvar/2)
        eps = torch.randn_like(std)
        return mu + eps * std
    def decode(self, z):
        return self.decoder(z)
    def forward(self, x,latent_vec=False):
        mu, logvar = self.encode(x)
        z = self.reparameterize(mu, logvar)
        x_reconst = self.decode(z)
        if latent_vec:
            return x_reconst, mu, logvar, z
        else:
            return x_reconst, mu, logvar
    def loss_function(self, x, x_hat, mu, logvar):
        recon_loss = calc_reconstruction_loss(x, x_hat, self.recon_loss_type)
        kl_loss = -0.5 * torch.mean(1 + logvar - mu**2 - logvar.exp())
        return kl_loss + recon_loss

class UpsamplingLayer(nn.Module):
    def __init__(self, in_channel, out_channel, activation="none", bn=True, type="transpose"):
        super(UpsamplingLayer, self).__init__()
        self.bn = nn.BatchNorm2d(out_channel) if bn else None
        if activation == "ReLU":
            self.activaton = nn.ReLU(True)
        elif activation == "none":
            self.activaton = None
        else:
            assert()
        if type == "transpose":
            self.upsampler = nn.Sequential(
                nn.ConvTranspose2d(in_channel, out_channel, kernel_size=2, stride=2, padding=0),
            )
        elif type == "upsample":
            self.upsampler = nn.Sequential(
                nn.UpsamplingBilinear2d(scale_factor=2),
                nn.Conv2d(in_channel, out_channel, kernel_size=3, stride=1, padding=1), 
            )
        else:
            assert()

    def forward(self, x):
        x = self.upsampler(x)
        if self.activaton:
            x = self.activaton(x)
        if self.bn:
            x = self.bn(x)
        return x

vqvae_model.py

import torch
from torch import nn
from torch.nn import functional as F
import abc

import pytorch_ssim
# Copyright 2018 The Sonnet Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or  implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================


# Borrowed from https://github.com/deepmind/sonnet and ported it to PyTorch

class AbstractAutoEncoder(nn.Module):
    __metaclass__ = abc.ABCMeta

    @abc.abstractmethod
    def encode(self, x):
        return

    @abc.abstractmethod
    def decode(self, z):
        return

    @abc.abstractmethod
    def forward(self, x, latent_vec=False):
        """model return (reconstructed_x, *)"""
        return

    @abc.abstractmethod
    def loss_function(self, **kwargs):
        """accepts (original images, *) where * is the same as returned from forward()"""
        return

    @abc.abstractmethod
    def latest_losses(self):
        """returns the latest losses in a dictionary. Useful for logging."""
        return


class Quantize(nn.Module):
    def __init__(self, dim, n_embed, decay=0.99, eps=1e-5):
        super().__init__()

        self.dim = dim
        self.n_embed = n_embed
        self.decay = decay
        self.eps = eps

        embed = torch.randn(dim, n_embed)
        self.register_buffer('embed', embed)
        self.register_buffer('cluster_size', torch.zeros(n_embed))
        self.register_buffer('embed_avg', embed.clone())

    def forward(self, input):
        flatten = input.reshape(-1, self.dim)
        dist = (
            flatten.pow(2).sum(1, keepdim=True)
            - 2 * flatten @ self.embed
            + self.embed.pow(2).sum(0, keepdim=True)
        )
        _, embed_ind = (-dist).max(1)
        embed_onehot = F.one_hot(embed_ind, self.n_embed).type(flatten.dtype)
        embed_ind = embed_ind.view(*input.shape[:-1])
        quantize = self.embed_code(embed_ind)

        if self.training:
            self.cluster_size.data.mul_(self.decay).add_(
                1 - self.decay, embed_onehot.sum(0)
            )
            embed_sum = flatten.transpose(0, 1) @ embed_onehot
            self.embed_avg.data.mul_(self.decay).add_(1 - self.decay, embed_sum)
            n = self.cluster_size.sum()
            cluster_size = (
                (self.cluster_size + self.eps) / (n + self.n_embed * self.eps) * n
            )
            embed_normalized = self.embed_avg / cluster_size.unsqueeze(0)
            self.embed.data.copy_(embed_normalized)

        diff = (quantize.detach() - input).pow(2).mean()
        quantize = input + (quantize - input).detach()

        return quantize, diff, embed_ind

    def embed_code(self, embed_id):
        return F.embedding(embed_id, self.embed.transpose(0, 1))


class ResBlock(nn.Module):
    def __init__(self, in_channel, channel):
        super().__init__()

        self.conv = nn.Sequential(
            nn.ReLU(inplace=True),
            nn.Conv2d(in_channel, channel, 3, padding=1),
            nn.ReLU(inplace=True),
            nn.Conv2d(channel, in_channel, 1),
        )

    def forward(self, input):
        out = self.conv(input)
        out += input

        return out


class Encoder(nn.Module):
    def __init__(self, in_channel, channel, n_res_block, n_res_channel, stride):
        super().__init__()

        if stride == 4:
            blocks = [
                nn.Conv2d(in_channel, channel // 2, 4, stride=2, padding=1),
                nn.ReLU(inplace=True),
                nn.Conv2d(channel // 2, channel, 4, stride=2, padding=1),
                nn.ReLU(inplace=True),
                nn.Conv2d(channel, channel, 3, padding=1),
            ]

        elif stride == 2:
            blocks = [
                nn.Conv2d(in_channel, channel // 2, 4, stride=2, padding=1),
                nn.ReLU(inplace=True),
                nn.Conv2d(channel // 2, channel, 3, padding=1),
            ]

        for i in range(n_res_block):
            blocks.append(ResBlock(channel, n_res_channel))

        blocks.append(nn.ReLU(inplace=True))

        self.blocks = nn.Sequential(*blocks)

    def forward(self, input):
        return self.blocks(input)


class Decoder(nn.Module):
    def __init__(
        self, in_channel, out_channel, channel, n_res_block, n_res_channel, stride
    ):
        super().__init__()

        blocks = [nn.Conv2d(in_channel, channel, 3, padding=1)]

        for i in range(n_res_block):
            blocks.append(ResBlock(channel, n_res_channel))

        blocks.append(nn.ReLU(inplace=True))

        if stride == 4:
            blocks.extend(
                [
                    nn.ConvTranspose2d(channel, channel // 2, 4, stride=2, padding=1),
                    nn.ReLU(inplace=True),
                    nn.ConvTranspose2d(
                        channel // 2, out_channel, 4, stride=2, padding=1
                    ),
                ]
            )

        elif stride == 2:
            blocks.append(
                nn.ConvTranspose2d(channel, out_channel, 4, stride=2, padding=1)
            )

        self.blocks = nn.Sequential(*blocks)

    def forward(self, input):
        return self.blocks(input)


class VQVAE(AbstractAutoEncoder):
    def __init__(
        self,
        recon_loss_type,
        in_channel=3,
        channel=128,
        n_res_block=2,
        n_res_channel=32,
        embed_dim=64,
        n_embed=512,
        decay=0.99,
    ):
        super().__init__()
        self.recon_loss_type = recon_loss_type
        self.enc_b = Encoder(in_channel, channel, n_res_block, n_res_channel, stride=4)
        self.enc_t = Encoder(channel, channel, n_res_block, n_res_channel, stride=2)
        self.quantize_conv_t = nn.Conv2d(channel, embed_dim, 1)
        self.quantize_t = Quantize(embed_dim, n_embed)
        self.dec_t = Decoder(
            embed_dim, embed_dim, channel, n_res_block, n_res_channel, stride=2
        )
        self.quantize_conv_b = nn.Conv2d(embed_dim + channel, embed_dim, 1)
        self.quantize_b = Quantize(embed_dim, n_embed)
        self.upsample_t = nn.ConvTranspose2d(
            embed_dim, embed_dim, 4, stride=2, padding=1
        )
        self.dec = Decoder(
            embed_dim + embed_dim,
            in_channel,
            channel,
            n_res_block,
            n_res_channel,
            stride=4,
        )

    def forward(self, input, latent_vec=False):
        quant_t, quant_b, diff, _, _ = self.encode(input)
        dec, quant = self.decode(quant_t, quant_b)
        if latent_vec:
            return dec, diff, quant
        else:
            return dec, diff

    def encode(self, input):
        enc_b = self.enc_b(input)
        enc_t = self.enc_t(enc_b)

        quant_t = self.quantize_conv_t(enc_t).permute(0, 2, 3, 1)
        quant_t, diff_t, id_t = self.quantize_t(quant_t)
        quant_t = quant_t.permute(0, 3, 1, 2)
        diff_t = diff_t.unsqueeze(0)

        dec_t = self.dec_t(quant_t)
        enc_b = torch.cat([dec_t, enc_b], 1)

        quant_b = self.quantize_conv_b(enc_b).permute(0, 2, 3, 1)
        quant_b, diff_b, id_b = self.quantize_b(quant_b)
        quant_b = quant_b.permute(0, 3, 1, 2)
        diff_b = diff_b.unsqueeze(0)

        return quant_t, quant_b, diff_t + diff_b, id_t, id_b

    def decode(self, quant_t, quant_b):
        upsample_t = self.upsample_t(quant_t)
        quant = torch.cat([upsample_t, quant_b], 1)
        dec = self.dec(quant)

        return dec, quant

    def decode_code(self, code_t, code_b):
        quant_t = self.quantize_t.embed_code(code_t)
        quant_t = quant_t.permute(0, 3, 1, 2)
        quant_b = self.quantize_b.embed_code(code_b)
        quant_b = quant_b.permute(0, 3, 1, 2)

        dec = self.decode(quant_t, quant_b)

        return dec

    def loss_function(self, x, x_hat, diff, latent_loss_weight=0.25):
        criterion = nn.MSELoss()
        recon_loss = criterion(x_hat, x)
        latent_loss = diff.mean()
        loss = recon_loss + latent_loss_weight * latent_loss
        return loss