github/DeepFaceLab
1
0
Fork 0
mirror of https://github.com/iperov/DeepFaceLab.git synced 2025-07-08 05:51:40 -07:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix avatar multigpu mode

Browse source
This commit is contained in:
iperov 2018-06-07 12:18:42 +04:00
parent adc1e701de
commit 58d9b4ae09
2 changed files with 37 additions and 29 deletions
Download patch file Download diff file Expand all files Collapse all files

45
models/Model_AVATAR/Model.py
View file

@ -3,6 +3,7 @@ from models import TrainingDataType
import numpy as np import numpy as np
import cv2 import cv2
from nnlib import tf_dssim from nnlib import tf_dssim
from nnlib import DSSIMLossClass
from nnlib import conv from nnlib import conv
from nnlib import upscale from nnlib import upscale
@ -37,42 +38,35 @@ class Model(ModelBase):
self.encoder64, self.decoder64_src, self.decoder64_dst, self.encoder256, self.decoder256 = self.to_multi_gpu_model_if_possible ( [self.encoder64, self.decoder64_src, self.decoder64_dst, self.encoder256, self.decoder256] ) self.encoder64, self.decoder64_src, self.decoder64_dst, self.encoder256, self.decoder256 = self.to_multi_gpu_model_if_possible ( [self.encoder64, self.decoder64_src, self.decoder64_dst, self.encoder256, self.decoder256] )
input_A_warped64 = keras.layers.Input(img_shape64) input_A_warped64 = keras.layers.Input(img_shape64)
input_A_target64 = keras.layers.Input(img_shape64)
input_B_warped64 = keras.layers.Input(img_shape64) input_B_warped64 = keras.layers.Input(img_shape64)
input_B_target64 = keras.layers.Input(img_shape64) A_rec64 = self.decoder64_src(self.encoder64(input_A_warped64))
B_rec64 = self.decoder64_dst(self.encoder64(input_B_warped64))
self.ae64 = self.keras.models.Model([input_A_warped64, input_B_warped64], [A_rec64, B_rec64] )
A_code64 = self.encoder64(input_A_warped64) if self.is_training_mode:
B_code64 = self.encoder64(input_B_warped64) self.ae64, = self.to_multi_gpu_model_if_possible ( [self.ae64,] )
A_rec64 = self.decoder64_src(A_code64) self.ae64.compile(optimizer=self.keras.optimizers.Adam(lr=5e-5, beta_1=0.5, beta_2=0.999),
B_rec64 = self.decoder64_dst(B_code64) loss=[DSSIMLossClass(self.tf)(), DSSIMLossClass(self.tf)()] )
A64_loss = tf_dssim(tf, input_A_target64, A_rec64)
B64_loss = tf_dssim(tf, input_B_target64, B_rec64)
total64_loss = A64_loss + B64_loss
self.ed64_train = K.function ([input_A_warped64, input_A_target64, input_B_warped64, input_B_target64],[K.mean(total64_loss)],
self.keras.optimizers.Adam(lr=5e-5, beta_1=0.5, beta_2=0.999).get_updates(total64_loss, self.encoder64.trainable_weights + self.decoder64_src.trainable_weights + self.decoder64_dst.trainable_weights)
)
self.A64_view = K.function ([input_A_warped64], [A_rec64]) self.A64_view = K.function ([input_A_warped64], [A_rec64])
self.B64_view = K.function ([input_B_warped64], [B_rec64]) self.B64_view = K.function ([input_B_warped64], [B_rec64])
input_A_warped64 = keras.layers.Input(img_shape64) input_A_warped64 = keras.layers.Input(img_shape64)
input_A_target256 = keras.layers.Input(img_shape256) input_A_target256 = keras.layers.Input(img_shape256)
A_code256 = self.encoder256(input_A_warped64) A_rec256 = self.decoder256( self.encoder256(input_A_warped64) )
A_rec256 = self.decoder256(A_code256)
input_B_warped64 = keras.layers.Input(img_shape64) input_B_warped64 = keras.layers.Input(img_shape64)
B_code64 = self.encoder64(input_B_warped64) BA_rec64 = self.decoder64_src( self.encoder64(input_B_warped64) )
BA_rec64 = self.decoder64_src(B_code64) BA_rec256 = self.decoder256( self.encoder256(BA_rec64) )
BA_code256 = self.encoder256(BA_rec64)
BA_rec256 = self.decoder256(BA_code256)
total256_loss = K.mean( tf_dssim(tf, input_A_target256, A_rec256) ) self.ae256 = self.keras.models.Model([input_A_warped64], [A_rec256] )
self.ed256_train = K.function ([input_A_warped64, input_A_target256],[total256_loss], if self.is_training_mode:
self.keras.optimizers.Adam(lr=5e-5, beta_1=0.5, beta_2=0.999).get_updates(total256_loss, self.encoder256.trainable_weights + self.decoder256.trainable_weights) self.ae256, = self.to_multi_gpu_model_if_possible ( [self.ae256,] )
)
self.ae256.compile(optimizer=self.keras.optimizers.Adam(lr=5e-5, beta_1=0.5, beta_2=0.999),
loss=[DSSIMLossClass(self.tf)()])
self.A256_view = K.function ([input_A_warped64], [A_rec256]) self.A256_view = K.function ([input_A_warped64], [A_rec256])
self.BA256_view = K.function ([input_B_warped64], [BA_rec256]) self.BA256_view = K.function ([input_B_warped64], [BA_rec256])
@ -108,8 +102,9 @@ class Model(ModelBase):
warped_src64, target_src64, target_src256, target_src_source64, target_src_source256 = sample[0] warped_src64, target_src64, target_src256, target_src_source64, target_src_source256 = sample[0]
warped_dst64, target_dst64, target_dst_source64, target_dst_source256 = sample[1] warped_dst64, target_dst64, target_dst_source64, target_dst_source256 = sample[1]
loss64, = self.ed64_train ([warped_src64, target_src64, warped_dst64, target_dst64]) loss64, loss_src64, loss_dst64 = self.ae64.train_on_batch ([warped_src64, warped_dst64], [target_src64, target_dst64])
loss256, = self.ed256_train ([warped_src64, target_src256])
loss256 = self.ae256.train_on_batch ([warped_src64], [target_src256])
return ( ('loss64', loss64 ), ('loss256', loss256), ) return ( ('loss64', loss64 ), ('loss256', loss256), )

13
nnlib/__init__.py
View file

@ -44,6 +44,19 @@ def DSSIMMaskLossClass(tf):
return DSSIMMaskLoss return DSSIMMaskLoss
def DSSIMLossClass(tf):
class DSSIMLoss(object):
def __init__(self, is_tanh=False):
self.is_tanh = is_tanh
def __call__(self,y_true, y_pred):
if not self.is_tanh:
return (1.0 - tf.image.ssim (y_true, y_pred, 1.0)) / 2.0
else:
return (1.0 - tf.image.ssim ((y_true/2+0.5), (y_pred/2+0.5), 1.0)) / 2.0
return DSSIMLoss
def MSEMaskLossClass(keras): def MSEMaskLossClass(keras):
class MSEMaskLoss(object): class MSEMaskLoss(object):
def __init__(self, mask_list, is_tanh=False): def __init__(self, mask_list, is_tanh=False):