Spaces:
Running
on
Zero
Running
on
Zero
File size: 11,452 Bytes
6efc863 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 |
import functools
import torch.nn as nn
class ActNorm(nn.Module):
def __init__(self, num_features, logdet=False, affine=True,
allow_reverse_init=False):
assert affine
super().__init__()
self.logdet = logdet
self.loc = nn.Parameter(torch.zeros(1, num_features, 1, 1))
self.scale = nn.Parameter(torch.ones(1, num_features, 1, 1))
self.allow_reverse_init = allow_reverse_init
self.register_buffer('initialized', torch.tensor(0, dtype=torch.uint8))
def initialize(self, input):
with torch.no_grad():
flatten = input.permute(1, 0, 2, 3).contiguous().view(input.shape[1], -1)
mean = (
flatten.mean(1)
.unsqueeze(1)
.unsqueeze(2)
.unsqueeze(3)
.permute(1, 0, 2, 3)
)
std = (
flatten.std(1)
.unsqueeze(1)
.unsqueeze(2)
.unsqueeze(3)
.permute(1, 0, 2, 3)
)
self.loc.data.copy_(-mean)
self.scale.data.copy_(1 / (std + 1e-6))
def forward(self, input, reverse=False):
if reverse:
return self.reverse(input)
if len(input.shape) == 2:
input = input[:, :, None, None]
squeeze = True
else:
squeeze = False
_, _, height, width = input.shape
if self.training and self.initialized.item() == 0:
self.initialize(input)
self.initialized.fill_(1)
h = self.scale * (input + self.loc)
if squeeze:
h = h.squeeze(-1).squeeze(-1)
if self.logdet:
log_abs = torch.log(torch.abs(self.scale))
logdet = height * width * torch.sum(log_abs)
logdet = logdet * torch.ones(input.shape[0]).to(input)
return h, logdet
return h
def reverse(self, output):
if self.training and self.initialized.item() == 0:
if not self.allow_reverse_init:
raise RuntimeError(
"Initializing ActNorm in reverse direction is "
"disabled by default. Use allow_reverse_init=True to enable."
)
else:
self.initialize(output)
self.initialized.fill_(1)
if len(output.shape) == 2:
output = output[:, :, None, None]
squeeze = True
else:
squeeze = False
h = output / self.scale - self.loc
if squeeze:
h = h.squeeze(-1).squeeze(-1)
return h
def weights_init(m):
classname = m.__class__.__name__
if classname.find('Conv') != -1:
nn.init.normal_(m.weight.data, 0.0, 0.02)
elif classname.find('BatchNorm') != -1:
nn.init.normal_(m.weight.data, 1.0, 0.02)
nn.init.constant_(m.bias.data, 0)
class NLayerDiscriminator(nn.Module):
"""Defines a PatchGAN discriminator as in Pix2Pix
--> see https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix/blob/master/models/networks.py
"""
def __init__(self, input_nc=3, ndf=64, n_layers=3, use_actnorm=False):
"""Construct a PatchGAN discriminator
Parameters:
input_nc (int) -- the number of channels in input images
ndf (int) -- the number of filters in the last conv layer
n_layers (int) -- the number of conv layers in the discriminator
norm_layer -- normalization layer
"""
super(NLayerDiscriminator, self).__init__()
if not use_actnorm:
norm_layer = nn.BatchNorm2d
else:
norm_layer = ActNorm
if type(norm_layer) == functools.partial: # no need to use bias as BatchNorm2d has affine parameters
use_bias = norm_layer.func != nn.BatchNorm2d
else:
use_bias = norm_layer != nn.BatchNorm2d
kw = 4
padw = 1
sequence = [nn.Conv2d(input_nc, ndf, kernel_size=kw, stride=2, padding=padw), nn.LeakyReLU(0.2, True)] # w/2
nf_mult = 1
nf_mult_prev = 1
for n in range(1, n_layers): # gradually increase the number of filters # w/(2**nlayers)
nf_mult_prev = nf_mult
nf_mult = min(2 ** n, 8)
sequence += [
nn.Conv2d(ndf * nf_mult_prev, ndf * nf_mult, kernel_size=kw, stride=2, padding=padw, bias=use_bias),
norm_layer(ndf * nf_mult),
nn.LeakyReLU(0.2, True)
]
nf_mult_prev = nf_mult
nf_mult = min(2 ** n_layers, 8)
sequence += [
nn.Conv2d(ndf * nf_mult_prev, ndf * nf_mult, kernel_size=kw, stride=1, padding=padw, bias=use_bias),# w - 1
norm_layer(ndf * nf_mult),
nn.LeakyReLU(0.2, True)
]
# output 1 channel prediction map
sequence += [nn.Conv2d(ndf * nf_mult, 1, kernel_size=kw, stride=1, padding=padw)] # w - 1
self.main = nn.Sequential(*sequence)
def forward(self, input):
"""Standard forward."""
return self.main(input)
class NLayerDiscriminator1dFeats(NLayerDiscriminator):
"""Defines a PatchGAN discriminator as in Pix2Pix
--> see https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix/blob/master/models/networks.py
"""
def __init__(self, input_nc=3, ndf=64, n_layers=3, use_actnorm=False):
"""Construct a PatchGAN discriminator
Parameters:
input_nc (int) -- the number of channels in input feats
ndf (int) -- the number of filters in the last conv layer
n_layers (int) -- the number of conv layers in the discriminator
norm_layer -- normalization layer
"""
super().__init__(input_nc=input_nc, ndf=64, n_layers=n_layers, use_actnorm=use_actnorm)
if not use_actnorm:
norm_layer = nn.BatchNorm1d
else:
norm_layer = ActNorm
if type(norm_layer) == functools.partial: # no need to use bias as BatchNorm has affine parameters
use_bias = norm_layer.func != nn.BatchNorm1d
else:
use_bias = norm_layer != nn.BatchNorm1d
kw = 4
padw = 1
sequence = [nn.Conv1d(input_nc, input_nc//2, kernel_size=kw, stride=2, padding=padw), nn.LeakyReLU(0.2, True)]
nf_mult = input_nc//2
nf_mult_prev = 1
for n in range(1, n_layers): # gradually decrease the number of filters
nf_mult_prev = nf_mult
nf_mult = max(nf_mult_prev // (2 ** n), 8)
sequence += [
nn.Conv1d(nf_mult_prev, nf_mult, kernel_size=kw, stride=2, padding=padw, bias=use_bias),
norm_layer(nf_mult),
nn.LeakyReLU(0.2, True)
]
nf_mult_prev = nf_mult
nf_mult = max(nf_mult_prev // (2 ** n), 8)
sequence += [
nn.Conv1d(nf_mult_prev, nf_mult, kernel_size=kw, stride=1, padding=padw, bias=use_bias),
norm_layer(nf_mult),
nn.LeakyReLU(0.2, True)
]
nf_mult_prev = nf_mult
nf_mult = max(nf_mult_prev // (2 ** n), 8)
sequence += [
nn.Conv1d(nf_mult_prev, nf_mult, kernel_size=kw, stride=1, padding=padw, bias=use_bias),
norm_layer(nf_mult),
nn.LeakyReLU(0.2, True)
]
# output 1 channel prediction map
sequence += [nn.Conv1d(nf_mult, 1, kernel_size=kw, stride=1, padding=padw)]
self.main = nn.Sequential(*sequence)
class NLayerDiscriminator1dSpecs(NLayerDiscriminator):
"""Defines a PatchGAN discriminator as in Pix2Pix
--> see https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix/blob/master/models/networks.py
"""
def __init__(self, input_nc=80, ndf=64, n_layers=3, use_actnorm=False):
"""Construct a PatchGAN discriminator
Parameters:
input_nc (int) -- the number of channels in input specs
ndf (int) -- the number of filters in the last conv layer
n_layers (int) -- the number of conv layers in the discriminator
norm_layer -- normalization layer
"""
super().__init__(input_nc=input_nc, ndf=64, n_layers=n_layers, use_actnorm=use_actnorm)
if not use_actnorm:
norm_layer = nn.BatchNorm1d
else:
norm_layer = ActNorm
if type(norm_layer) == functools.partial: # no need to use bias as BatchNorm has affine parameters
use_bias = norm_layer.func != nn.BatchNorm1d
else:
use_bias = norm_layer != nn.BatchNorm1d
kw = 4
padw = 1
sequence = [nn.Conv1d(input_nc, ndf, kernel_size=kw, stride=2, padding=padw), nn.LeakyReLU(0.2, True)]
nf_mult = 1
nf_mult_prev = 1
for n in range(1, n_layers): # gradually decrease the number of filters
nf_mult_prev = nf_mult
nf_mult = min(2 ** n, 8)
sequence += [
nn.Conv1d(ndf * nf_mult_prev, ndf * nf_mult, kernel_size=kw, stride=2, padding=padw, bias=use_bias),
norm_layer(ndf * nf_mult),
nn.LeakyReLU(0.2, True)
]
nf_mult_prev = nf_mult
nf_mult = min(2 ** n_layers, 8)
sequence += [
nn.Conv1d(ndf * nf_mult_prev, ndf * nf_mult, kernel_size=kw, stride=1, padding=padw, bias=use_bias),
norm_layer(ndf * nf_mult),
nn.LeakyReLU(0.2, True)
]
# output 1 channel prediction map
sequence += [nn.Conv1d(ndf * nf_mult, 1, kernel_size=kw, stride=1, padding=padw)]
self.main = nn.Sequential(*sequence)
def forward(self, input):
"""Standard forward."""
# (B, C, L)
input = input.squeeze(1)
input = self.main(input)
return input
if __name__ == '__main__':
import torch
## FEATURES
disc_in_channels = 2048
disc_num_layers = 2
use_actnorm = False
disc_ndf = 64
discriminator = NLayerDiscriminator1dFeats(input_nc=disc_in_channels, n_layers=disc_num_layers,
use_actnorm=use_actnorm, ndf=disc_ndf).apply(weights_init)
inputs = torch.rand((6, 2048, 212))
outputs = discriminator(inputs)
print(outputs.shape)
## AUDIO
disc_in_channels = 1
disc_num_layers = 3
use_actnorm = False
disc_ndf = 64
discriminator = NLayerDiscriminator(input_nc=disc_in_channels, n_layers=disc_num_layers,
use_actnorm=use_actnorm, ndf=disc_ndf).apply(weights_init)
inputs = torch.rand((6, 1, 80, 848))
outputs = discriminator(inputs)
print(outputs.shape)
## IMAGE
disc_in_channels = 3
disc_num_layers = 3
use_actnorm = False
disc_ndf = 64
discriminator = NLayerDiscriminator(input_nc=disc_in_channels, n_layers=disc_num_layers,
use_actnorm=use_actnorm, ndf=disc_ndf).apply(weights_init)
inputs = torch.rand((6, 3, 256, 256))
outputs = discriminator(inputs)
print(outputs.shape)
|