Spaces:
Runtime error
Runtime error
File size: 4,535 Bytes
9994352 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 |
import torch
import torch.nn as nn
import torch.nn.functional as F
import numpy as np
"""Class to create Residual block."""
class ResidualBlock(nn.Module):
def __init__(self, channels: int):
super(ResidualBlock, self).__init__()
self.conv_1 = nn.Conv2d(channels, channels, kernel_size = 3, stride = 1, padding = 1, padding_mode = 'reflect')
self.inst_1 = nn.InstanceNorm2d(channels, affine = True)
self.conv_2 = nn.Conv2d(channels, channels, kernel_size = 3, stride = 1, padding = 1, padding_mode = 'reflect')
self.inst_2 = nn.InstanceNorm2d(channels, affine = True)
self.relu = nn.ReLU()
def forward(self, x: torch.Tensor) -> torch.Tensor:
residue = x
out = self.relu(self.inst_1(self.conv_1(x)))
out = self.inst_2(self.conv_2(out))
return residue + out
"""_____________________________________________________________________________________________________________________________________________________________"""
"""Class to create Upsampling of image."""
class UpConv(nn.Module):
def __init__(self, in_channels: int, out_channels: int, kernel_size: int, stride: int, padding: int):
super(UpConv, self).__init__()
self.factor = stride
self.conv = nn.Conv2d(in_channels, out_channels, kernel_size = kernel_size,
stride = 1, padding = padding, padding_mode = 'reflect'
)
def forward(self, x: torch.Tensor) -> torch.Tensor:
if self.factor > 1:
x = F.interpolate(x, scale_factor = self.factor)
return self.conv(x)
"""_____________________________________________________________________________________________________________________________________________________________"""
"""Class to create Transformer Net."""
class TransFormerNet(nn.Module):
"""
Leon A. Gatys, Alexander S. Ecker, Matthias Bethge paper [1].
Justin Johnson, Alexandre Alahi, Li Fei-Fei paper [2].
It follows Original Johnson's Architecture [3].
Instance normalization is used in the place of Batch Normalization. It prevents instance-specific mean and covariance shift simplifying the learning process [4].
"""
def __init__(self):
super().__init__()
# Down Conv
self.conv1 = nn.Conv2d(3, 32, kernel_size = 9, stride = 1, padding = 9 // 2, padding_mode = 'reflect')
self.inst1 = nn.InstanceNorm2d(32, affine = True)
self.conv2 = nn.Conv2d(32, 64, kernel_size = 3, stride = 2, padding = 1, padding_mode = 'reflect')
self.inst2 = nn.InstanceNorm2d(64, affine = True)
self.conv3 = nn.Conv2d(64, 128, kernel_size = 3, stride = 2, padding = 1, padding_mode = 'reflect')
self.inst3 = nn.InstanceNorm2d(128, affine = True)
# Residual Blocks
self.resblock1 = ResidualBlock(128)
self.resblock2 = ResidualBlock(128)
self.resblock3 = ResidualBlock(128)
self.resblock4 = ResidualBlock(128)
self.resblock5 = ResidualBlock(128)
# Up conv
self.up_conv1 = UpConv(128, 64, kernel_size = 3, stride = 2, padding = 1)
self.inst4 = nn.InstanceNorm2d(64, affine = True)
self.up_conv2 = UpConv(64, 32, kernel_size = 3, stride = 2, padding = 1)
self.inst5 = nn.InstanceNorm2d(32, affine = True)
self.up_conv3 = UpConv(32, 3, kernel_size = 9, stride = 1, padding = 9 // 2)
def forward(self, x: torch.Tensor) -> torch.Tensor:
out = F.relu(self.inst1(self.conv1(x)))
out = F.relu(self.inst2(self.conv2(out)))
out = F.relu(self.inst3(self.conv3(out)))
out = self.resblock1(out)
out = self.resblock2(out)
out = self.resblock3(out)
out = self.resblock4(out)
out = self.resblock5(out)
out = F.relu(self.inst4(self.up_conv1(out)))
out = F.relu(self.inst5(self.up_conv2(out)))
out = self.up_conv3(out)
return out
if __name__ == '__main__':
net = TransFormerNet()
Num_of_parameters = sum(p.numel() for p in net.parameters())
print("Model Parameters : {:.3f} M".format(Num_of_parameters / 1e6))
"""
References:
[1] https://arxiv.org/abs/1508.06576
[2] https://arxiv.org/abs/1603.08155
[3] https://cs.stanford.edu/people/jcjohns/papers/fast-style/fast-style-supp.pdf
[4] https://arxiv.org/pdf/1607.08022.pdf
""" |