File size: 15,136 Bytes
b53f252 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 |
import torch
import torch.nn as nn
import torch.nn.parallel
from miscc.config import cfg
from torch.autograd import Variable
def conv3x1(in_planes, out_planes, stride=1):
"3x1 convolution with padding"
kernel_length = 41
return nn.Conv1d(in_planes, out_planes, kernel_size=kernel_length, stride=stride,
padding=20, bias=False)
def old_conv3x1(in_planes, out_planes, stride=1):
"3x1 convolution with padding"
kernel_length = 3
return nn.Conv1d(in_planes, out_planes, kernel_size=kernel_length, stride=stride,
padding=1, bias=False)
# def convn3x1(in_planes, out_planes, stride=1):
# "3x1 convolution with padding"
# return nn.Conv1d(in_planes, out_planes, kernel_size=9, stride=stride,
# padding=4, bias=False)
# Upsale the spatial size by a factor of 2
def upBlock4(in_planes, out_planes):
kernel_length = 41
stride = 4
block = nn.Sequential(
# nn.Upsample(scale_factor=4, mode='nearest'),
# conv3x1(in_planes, out_planes),
nn.ConvTranspose1d(in_planes,out_planes,kernel_size=kernel_length,stride=stride, padding=19,output_padding=1),
nn.BatchNorm1d(out_planes),
# nn.ReLU(True)
nn.PReLU())
return block
def upBlock2(in_planes, out_planes):
kernel_length = 41
stride = 2
block = nn.Sequential(
# nn.Upsample(scale_factor=4, mode='nearest'),
# conv3x1(in_planes, out_planes),
nn.ConvTranspose1d(in_planes,out_planes,kernel_size=kernel_length,stride=stride, padding=20,output_padding=1),
nn.BatchNorm1d(out_planes),
# nn.ReLU(True)
nn.PReLU())
return block
def sameBlock(in_planes, out_planes):
block = nn.Sequential(
# nn.Upsample(scale_factor=4, mode='nearest'),
conv3x1(in_planes, out_planes),
nn.BatchNorm1d(out_planes),
# nn.ReLU(True)
nn.PReLU())
return block
class ResBlock(nn.Module):
def __init__(self, channel_num):
super(ResBlock, self).__init__()
self.block = nn.Sequential(
conv3x1(channel_num, channel_num),
nn.BatchNorm1d(channel_num),
# nn.ReLU(True),
nn.PReLU(),
conv3x1(channel_num, channel_num),
nn.BatchNorm1d(channel_num))
self.relu = nn.PReLU()#nn.ReLU(inplace=True)
def forward(self, x):
residual = x
out = self.block(x)
out += residual
out = self.relu(out)
return out
# class CA_NET(nn.Module): #not chnaged yet
# # some code is modified from vae examples
# # (https://github.com/pytorch/examples/blob/master/vae/main.py)
# def __init__(self):
# super(CA_NET, self).__init__()
# self.t_dim = cfg.TEXT.DIMENSION
# self.c_dim = cfg.GAN.CONDITION_DIM
# self.fc = nn.Linear(self.t_dim, self.c_dim * 2, bias=True)
# self.relu = nn.ReLU()
# def encode(self, text_embedding):
# x = self.relu(self.fc(text_embedding))
# mu = x[:, :self.c_dim]
# logvar = x[:, self.c_dim:]
# return mu, logvar
# def reparametrize(self, mu, logvar):
# std = logvar.mul(0.5).exp_()
# if cfg.CUDA:
# eps = torch.cuda.FloatTensor(std.size()).normal_()
# else:
# eps = torch.FloatTensor(std.size()).normal_()
# eps = Variable(eps)
# return eps.mul(std).add_(mu)
# def forward(self, text_embedding):
# mu, logvar = self.encode(text_embedding)
# c_code = self.reparametrize(mu, logvar)
# return c_code, mu, logvar
class COND_NET(nn.Module): #not chnaged yet
# some code is modified from vae examples
# (https://github.com/pytorch/examples/blob/master/vae/main.py)
def __init__(self):
super(COND_NET, self).__init__()
self.t_dim = cfg.TEXT.DIMENSION
self.c_dim = cfg.GAN.CONDITION_DIM
self.fc = nn.Linear(self.t_dim, self.c_dim, bias=True)
self.relu = nn.PReLU()#nn.ReLU()
def encode(self, text_embedding):
x = self.relu(self.fc(text_embedding))
# mu = x[:, :self.c_dim]
# logvar = x[:, self.c_dim:]
return x
# def reparametrize(self, mu, logvar):
# std = logvar.mul(0.5).exp_()
# if cfg.CUDA:
# eps = torch.cuda.FloatTensor(std.size()).normal_()
# else:
# eps = torch.FloatTensor(std.size()).normal_()
# eps = Variable(eps)
# return eps.mul(std).add_(mu)
def forward(self, text_embedding):
c_code = self.encode(text_embedding)
# c_code = self.reparametrize(mu, logvar)
return c_code #, mu, logvar
class D_GET_LOGITS(nn.Module): #not chnaged yet
def __init__(self, ndf, nef, bcondition=True):
super(D_GET_LOGITS, self).__init__()
self.df_dim = ndf
self.ef_dim = nef
self.bcondition = bcondition
kernel_length =41
if bcondition:
self.convd1d = nn.ConvTranspose1d(ndf*8,ndf //2,kernel_size=kernel_length,stride=1, padding=20)
# self.outlogits = nn.Sequential(
# old_conv3x1(ndf * 8 + nef, ndf * 8),
# nn.BatchNorm1d(ndf * 8),
# nn.LeakyReLU(0.2, inplace=True),
# nn.Conv1d(ndf * 8, 1, kernel_size=16, stride=4),
# # nn.Conv1d(1, 1, kernel_size=16, stride=4),
# nn.Sigmoid()
# )
self.outlogits = nn.Sequential(
old_conv3x1(ndf //2 + nef, ndf //2 ),
nn.BatchNorm1d(ndf //2 ),
nn.LeakyReLU(0.2, inplace=True),
nn.Conv1d(ndf //2 , 1, kernel_size=16, stride=4),
# nn.Conv1d(1, 1, kernel_size=16, stride=4),
nn.Sigmoid()
)
else:
# self.outlogits = nn.Sequential(
# nn.Conv1d(ndf * 8, 1, kernel_size=16, stride=4),
# # nn.Conv1d(1, 1, kernel_size=16, stride=4),
# nn.Sigmoid())
self.convd1d = nn.ConvTranspose1d(ndf*8,ndf //2,kernel_size=kernel_length,stride=1, padding=20)
self.outlogits = nn.Sequential(
nn.Conv1d(ndf // 2 , 1, kernel_size=16, stride=4),
# nn.Conv1d(1, 1, kernel_size=16, stride=4),
nn.Sigmoid())
def forward(self, h_code, c_code=None):
# conditioning output
h_code = self.convd1d(h_code)
if self.bcondition and c_code is not None:
#print("mode c_code1 ",c_code.size())
c_code = c_code.view(-1, self.ef_dim, 1)
#print("mode c_code2 ",c_code.size())
c_code = c_code.repeat(1, 1, 16)
# state size (ngf+egf) x 16
#print("mode c_code ",c_code.size())
#print("mode h_code ",h_code.size())
h_c_code = torch.cat((h_code, c_code), 1)
else:
h_c_code = h_code
output = self.outlogits(h_c_code)
return output.view(-1)
# ############# Networks for stageI GAN #############
class STAGE1_G(nn.Module):
def __init__(self):
super(STAGE1_G, self).__init__()
self.gf_dim = cfg.GAN.GF_DIM * 8
self.ef_dim = cfg.GAN.CONDITION_DIM
# self.z_dim = cfg.Z_DIM
self.define_module()
def define_module(self):
kernel_length = 41
ninput = self.ef_dim #self.z_dim + self.ef_dim
ngf = self.gf_dim
# TEXT.DIMENSION -> GAN.CONDITION_DIM
# self.ca_net = CA_NET()
self.cond_net = COND_NET()
# -> ngf x 16
self.fc = nn.Sequential(
nn.Linear(ninput, ngf * 16, bias=False),
nn.BatchNorm1d(ngf * 16),
# nn.ReLU(True)
nn.PReLU())
# ngf x 16 -> ngf/2 x 64
self.upsample1 = upBlock4(ngf, ngf // 2)
# -> ngf/4 x 256
self.upsample2 = upBlock4(ngf // 2, ngf // 4)
# -> ngf/8 x 1024
self.upsample3 = upBlock4(ngf // 4, ngf // 8)
# -> ngf/16 x 4096
self.upsample4 = upBlock2(ngf // 8, ngf // 16)
self.upsample5 = upBlock2(ngf // 16, ngf // 16)
# -> 1 x 4096
self.RIR = nn.Sequential(
nn.ConvTranspose1d(ngf // 16,1,kernel_size=kernel_length,stride=1, padding=20),
# old_conv3x1(ngf // 16, 1), # conv3x3(ngf // 16, 3),
nn.Tanh())
def forward(self, text_embedding):
# c_code, mu, logvar = self.ca_net(text_embedding)
c_code = self.cond_net(text_embedding)
# z_c_code = torch.cat((noise, c_code), 1)
h_code = self.fc(c_code)
h_code = h_code.view(-1, self.gf_dim, 16)
# #print("h_code 1 ",h_code.size())
h_code = self.upsample1(h_code)
# #print("h_code 2 ",h_code.size())
h_code = self.upsample2(h_code)
# #print("h_code 3 ",h_code.size())
h_code = self.upsample3(h_code)
# #print("h_code 4 ",h_code.size())
h_code = self.upsample4(h_code)
h_code = self.upsample5(h_code)
# #print("h_code 5 ",h_code.size())
# state size 3 x 64 x 64
fake_RIR = self.RIR(h_code)
# return None, fake_RIR, mu, logvar
#print("generator ", text_embedding.size())
return None, fake_RIR, text_embedding #c_code
class STAGE1_D(nn.Module):
def __init__(self):
super(STAGE1_D, self).__init__()
self.df_dim = cfg.GAN.DF_DIM
self.ef_dim = cfg.GAN.CONDITION_DIM
self.define_module()
def define_module(self):
ndf, nef = self.df_dim, self.ef_dim
kernel_length =41
self.encode_RIR = nn.Sequential(
nn.Conv1d(1, ndf, kernel_length, 4, 20, bias=False),
nn.LeakyReLU(0.2, inplace=True),
# state size. (ndf) x 1024
nn.Conv1d(ndf, ndf * 2, kernel_length, 4, 20, bias=False),
nn.BatchNorm1d(ndf * 2),
nn.LeakyReLU(0.2, inplace=True),
# state size (ndf*2) x 256
nn.Conv1d(ndf*2, ndf * 4, kernel_length, 4, 20, bias=False),
nn.BatchNorm1d(ndf * 4),
nn.LeakyReLU(0.2, inplace=True),
# # state size (ndf*4) x 64
nn.Conv1d(ndf*4, ndf * 8, kernel_length, 4, 20, bias=False),
nn.BatchNorm1d(ndf * 8),
# state size (ndf * 8) x 16)
nn.LeakyReLU(0.2, inplace=True)
)
self.get_cond_logits = D_GET_LOGITS(ndf, nef)
self.get_uncond_logits = None
def forward(self, RIRs):
#print("model RIRs ",RIRs.size())
RIR_embedding = self.encode_RIR(RIRs)
#print("models RIR_embedding ",RIR_embedding.size())
return RIR_embedding
# ############# Networks for stageII GAN #############
class STAGE2_G(nn.Module):
def __init__(self, STAGE1_G):
super(STAGE2_G, self).__init__()
self.gf_dim = cfg.GAN.GF_DIM
self.ef_dim = cfg.GAN.CONDITION_DIM
# self.z_dim = cfg.Z_DIM
self.STAGE1_G = STAGE1_G
# fix parameters of stageI GAN
for param in self.STAGE1_G.parameters():
param.requires_grad = False
self.define_module()
def _make_layer(self, block, channel_num):
layers = []
for i in range(cfg.GAN.R_NUM):
layers.append(block(channel_num))
return nn.Sequential(*layers)
def define_module(self):
ngf = self.gf_dim
# TEXT.DIMENSION -> GAN.CONDITION_DIM
# self.ca_net = CA_NET()
self.cond_net = COND_NET()
# --> 4ngf x 16 x 16
self.encoder = nn.Sequential(
conv3x1(1, ngf),
nn.ReLU(True),
nn.Conv1d(ngf, ngf * 2, 16, 4, 6, bias=False),
nn.BatchNorm1d(ngf * 2),
nn.ReLU(True),
nn.Conv1d(ngf * 2, ngf * 4, 16, 4, 6, bias=False),
nn.BatchNorm1d(ngf * 4),
nn.ReLU(True))
self.hr_joint = nn.Sequential(
conv3x1(self.ef_dim + ngf * 4, ngf * 4),
nn.BatchNorm1d(ngf * 4),
nn.ReLU(True))
self.residual = self._make_layer(ResBlock, ngf * 4)
# --> 2ngf x 1024
self.upsample1 = upBlock4(ngf * 4, ngf * 2)
# --> ngf x 4096
self.upsample2 = upBlock4(ngf * 2, ngf)
# --> ngf // 2 x 16384
self.upsample3 = upBlock4(ngf, ngf // 2)
# --> ngf // 4 x 16384
self.upsample4 = sameBlock(ngf // 2, ngf // 4)
# --> 1 x 16384
self.RIR = nn.Sequential(
conv3x1(ngf // 4, 1),
nn.Tanh())
def forward(self, text_embedding):
_, stage1_RIR, _= self.STAGE1_G(text_embedding)
stage1_RIR = stage1_RIR.detach()
encoded_RIR = self.encoder(stage1_RIR)
# c_code, mu, logvar = self.ca_net(text_embedding)
c_code1 = self.cond_net(text_embedding)
c_code = c_code1.view(-1, self.ef_dim, 1)
c_code = c_code.repeat(1, 1, 256) # c_code.repeat(1, 1, 16, 16)
i_c_code = torch.cat([encoded_RIR, c_code], 1)
h_code = self.hr_joint(i_c_code)
h_code = self.residual(h_code)
h_code = self.upsample1(h_code)
h_code = self.upsample2(h_code)
h_code = self.upsample3(h_code)
h_code = self.upsample4(h_code)
fake_RIR = self.RIR(h_code)
return stage1_RIR, fake_RIR, c_code1 #mu, logvar
class STAGE2_D(nn.Module):
def __init__(self):
super(STAGE2_D, self).__init__()
self.df_dim = cfg.GAN.DF_DIM
self.ef_dim = cfg.GAN.CONDITION_DIM
self.define_module()
def define_module(self):
ndf, nef = self.df_dim, self.ef_dim
self.encode_RIR = nn.Sequential(
nn.Conv1d(1, ndf, 3, 1, 1, bias=False), # 16384 * ndf
nn.LeakyReLU(0.2, inplace=True),
nn.Conv1d(ndf, ndf * 2, 16, 4, 6, bias=False),
nn.BatchNorm1d(ndf * 2),
nn.LeakyReLU(0.2, inplace=True), # 4096 * ndf * 2
nn.Conv1d(ndf * 2, ndf * 4, 16, 4, 6, bias=False),
nn.BatchNorm1d(ndf * 4),
nn.LeakyReLU(0.2, inplace=True), # 1024 * ndf * 4
nn.Conv1d(ndf * 4, ndf * 8, 16, 4, 6, bias=False),
nn.BatchNorm1d(ndf * 8),
nn.LeakyReLU(0.2, inplace=True), # 256 * ndf * 8
nn.Conv1d(ndf * 8, ndf * 16, 16, 4, 6, bias=False),
nn.BatchNorm1d(ndf * 16),
nn.LeakyReLU(0.2, inplace=True), # 64 * ndf * 16
nn.Conv1d(ndf * 16, ndf * 32, 16, 4, 6, bias=False),
nn.BatchNorm1d(ndf * 32),
nn.LeakyReLU(0.2, inplace=True), # 16 * ndf * 32
conv3x1(ndf * 32, ndf * 16),
nn.BatchNorm1d(ndf * 16),
nn.LeakyReLU(0.2, inplace=True), # 16 * ndf * 16
conv3x1(ndf * 16, ndf * 8),
nn.BatchNorm1d(ndf * 8),
nn.LeakyReLU(0.2, inplace=True) # 16 * ndf * 8
)
self.get_cond_logits = D_GET_LOGITS(ndf, nef, bcondition=True)
self.get_uncond_logits = D_GET_LOGITS(ndf, nef, bcondition=False)
def forward(self, RIRs):
RIR_embedding = self.encode_RIR(RIRs)
return RIR_embedding
|