CalvinBackup/facefusion-labs
1
0
Fork 0
mirror of https://github.com/facefusion/facefusion-labs.git synced 2026-05-22 23:59:40 +02:00
Code Issues Packages Projects Releases Wiki Activity

Let the generator return target attributes

Browse Source
This commit is contained in:
henryruhs
2025-03-12 15:33:31 +01:00
parent 738e00d59e
commit 569df9e96d
2 changed files with 10 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files
face_swapper/src/models/generator.py
+3 -6
View File
@@ -16,13 +16,10 @@ class Generator(nn.Module):
self.encoder.apply(init_weight)
self.generator.apply(init_weight)
def forward(self, source_embedding : Embedding, target_tensor : Tensor) -> Tensor:
target_attributes = self.get_attributes(target_tensor)
def forward(self, source_embedding : Embedding, target_tensor : Tensor) -> Tuple[Tensor, Tuple[Attribute, ...]]:
target_attributes = self.encoder(target_tensor)
output_tensor = self.generator(source_embedding, target_attributes)
return output_tensor
def get_attributes(self, input_tensor : Tensor) -> Tuple[Attribute, ...]:
return self.encoder(input_tensor)
return output_tensor, target_attributes
def init_weight(module : nn.Module) -> None:
face_swapper/src/training.py
+7 -8
View File
@@ -55,8 +55,8 @@ class FaceSwapperTrainer(LightningModule):
def forward(self, source_embedding : Embedding, target_tensor : Tensor) -> Tuple[Tensor, Tensor]:
with torch.no_grad():
output_tensor = self.generator(source_embedding, target_tensor)
target_attribute = self.generator.get_attributes(target_tensor)[-1]
output_tensor, target_attributes = self.generator(source_embedding, target_tensor)
target_attribute = target_attributes[-1]
mask_tensor = self.masker(target_tensor, target_attribute)
return output_tensor, mask_tensor
@@ -105,12 +105,11 @@ class FaceSwapperTrainer(LightningModule):
self.toggle_optimizer(generator_optimizer)
source_embedding = calc_embedding(self.embedder, source_tensor, (0, 0, 0, 0))
target_attributes = self.generator.get_attributes(target_tensor)
generator_output_tensor = self.generator(source_embedding, target_tensor)
generator_output_attributes = self.generator.get_attributes(generator_output_tensor)
_, generator_target_attributes = self.generator(target_tensor)
generator_output_tensor, generator_output_attributes = self.generator(source_embedding, target_tensor)
discriminator_output_tensors = self.discriminator(generator_output_tensor)
adversarial_loss, weighted_adversarial_loss = self.adversarial_loss(discriminator_output_tensors)
attribute_loss, weighted_attribute_loss = self.attribute_loss(target_attributes, generator_output_attributes)
attribute_loss, weighted_attribute_loss = self.attribute_loss(generator_target_attributes, generator_output_attributes)
reconstruction_loss, weighted_reconstruction_loss = self.reconstruction_loss(source_tensor, target_tensor, generator_output_tensor)
identity_loss, weighted_identity_loss = self.identity_loss(generator_output_tensor, source_tensor)
pose_loss, weighted_pose_loss, expression_loss, weighted_expression_loss = self.motion_loss(target_tensor, generator_output_tensor)
@@ -126,7 +125,7 @@ class FaceSwapperTrainer(LightningModule):
self.untoggle_optimizer(generator_optimizer)
self.toggle_optimizer(masker_optimizer)
target_attribute = target_attributes[-1].detach()
target_attribute = generator_target_attributes[-1].detach()
mask_tensor = self.masker(target_tensor, target_attribute)
mask_loss = self.mask_loss(target_tensor, mask_tensor)
@@ -168,7 +167,7 @@ class FaceSwapperTrainer(LightningModule):
def validation_step(self, batch : Batch, batch_index : int) -> Tensor:
source_tensor, target_tensor = batch
source_embedding = calc_embedding(self.embedder, source_tensor, (0, 0, 0, 0))
output_tensor = self.generator(source_embedding, target_tensor)
output_tensor, _ = self.generator(source_embedding, target_tensor)
output_embedding = calc_embedding(self.embedder, output_tensor, (0, 0, 0, 0))
validation_score = (nn.functional.cosine_similarity(source_embedding, output_embedding).mean() + 1) * 0.5
self.log('validation_score', validation_score, prog_bar = True)