mirror of
https://github.com/facefusion/facefusion-labs.git
synced 2026-04-19 15:56:37 +02:00
harisreedhar
206 lines
8.9 KiB
Python
206 lines
8.9 KiB
Python
import configparser
|
|
import os
|
|
from typing import Tuple
|
|
|
|
import lightning
|
|
import torch
|
|
import torchvision
|
|
from lightning import Trainer
|
|
from lightning.pytorch.callbacks import ModelCheckpoint
|
|
from lightning.pytorch.loggers import TensorBoardLogger
|
|
from torch import Tensor, nn
|
|
from torch.utils.data import Dataset, random_split
|
|
from torchdata.stateful_dataloader import StatefulDataLoader
|
|
|
|
from .dataset import DynamicDataset
|
|
from .helper import calc_embedding
|
|
from .models.discriminator import Discriminator
|
|
from .models.generator import Generator
|
|
from .models.loss import AdversarialLoss, AttributeLoss, DiscriminatorLoss, GazeLoss, IdentityLoss, PoseLoss, ReconstructionLoss
|
|
from .types import Batch, Embedding, OptimizerConfig
|
|
|
|
CONFIG = configparser.ConfigParser()
|
|
CONFIG.read('config.ini')
|
|
|
|
|
|
class FaceSwapperTrainer(lightning.LightningModule):
|
|
def __init__(self) -> None:
|
|
super().__init__()
|
|
embedder_path = CONFIG.get('training.model', 'embedder_path')
|
|
|
|
self.generator = Generator()
|
|
self.discriminator = Discriminator()
|
|
self.discriminator_loss = DiscriminatorLoss()
|
|
self.adversarial_loss = AdversarialLoss()
|
|
self.attribute_loss = AttributeLoss()
|
|
self.reconstruction_loss = ReconstructionLoss()
|
|
self.identity_loss = IdentityLoss()
|
|
self.pose_loss = PoseLoss()
|
|
self.gaze_loss = GazeLoss()
|
|
self.embedder = torch.jit.load(embedder_path, map_location = 'cpu') # type:ignore[no-untyped-call]
|
|
self.automatic_optimization = False
|
|
|
|
def forward(self, target_tensor : Tensor, source_embedding : Embedding) -> Tensor:
|
|
output_tensor = self.generator(source_embedding, target_tensor)
|
|
return output_tensor
|
|
|
|
def configure_optimizers(self) -> Tuple[OptimizerConfig, OptimizerConfig]:
|
|
learning_rate = CONFIG.getfloat('training.trainer', 'learning_rate')
|
|
generator_optimizer = torch.optim.Adam(self.generator.parameters(), lr = learning_rate, betas = (0.0, 0.999), weight_decay = 1e-4)
|
|
discriminator_optimizer = torch.optim.Adam(self.discriminator.parameters(), lr = learning_rate, betas = (0.0, 0.999), weight_decay = 1e-4)
|
|
generator_scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(generator_optimizer, T_max = 10, eta_min = 1e-6)
|
|
discriminator_scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(discriminator_optimizer, T_max = 10, eta_min = 1e-6)
|
|
|
|
generator_config =\
|
|
{
|
|
'optimizer': generator_optimizer,
|
|
'lr_scheduler':
|
|
{
|
|
'scheduler': generator_scheduler,
|
|
'interval': 'step',
|
|
}
|
|
}
|
|
discriminator_config =\
|
|
{
|
|
'optimizer': discriminator_optimizer,
|
|
'lr_scheduler':
|
|
{
|
|
'scheduler': discriminator_scheduler,
|
|
'interval': 'step',
|
|
}
|
|
}
|
|
return generator_config, discriminator_config
|
|
|
|
def training_step(self, batch : Batch, batch_index : int) -> Tensor:
|
|
preview_frequency = CONFIG.getfloat('training.trainer', 'preview_frequency')
|
|
|
|
source_tensor, target_tensor = batch
|
|
generator_optimizer, discriminator_optimizer = self.optimizers() #type:ignore[attr-defined]
|
|
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)
|
|
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)
|
|
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 = self.pose_loss(target_tensor, generator_output_tensor)
|
|
gaze_loss, weighted_gaze_loss = self.gaze_loss(target_tensor, generator_output_tensor)
|
|
generator_loss = weighted_adversarial_loss + weighted_attribute_loss + weighted_reconstruction_loss + weighted_identity_loss + weighted_pose_loss + weighted_gaze_loss
|
|
|
|
generator_optimizer.zero_grad()
|
|
self.manual_backward(generator_loss)
|
|
generator_optimizer.step()
|
|
|
|
generator_scheduler = self.lr_schedulers()[0]
|
|
generator_scheduler.step()
|
|
|
|
discriminator_source_tensors = self.discriminator(source_tensor)
|
|
discriminator_output_tensors = self.discriminator(generator_output_tensor.detach())
|
|
discriminator_loss = self.discriminator_loss(discriminator_source_tensors, discriminator_output_tensors)
|
|
|
|
discriminator_optimizer.zero_grad()
|
|
self.manual_backward(discriminator_loss)
|
|
discriminator_optimizer.step()
|
|
|
|
discriminator_scheduler = self.lr_schedulers()[1]
|
|
discriminator_scheduler.step()
|
|
|
|
if self.global_step % preview_frequency == 0:
|
|
self.generate_preview(source_tensor, target_tensor, generator_output_tensor)
|
|
|
|
self.log('generator_loss', generator_loss, prog_bar = True)
|
|
self.log('discriminator_loss', discriminator_loss, prog_bar = True)
|
|
self.log('adversarial_loss', adversarial_loss)
|
|
self.log('attribute_loss', attribute_loss)
|
|
self.log('reconstruction_loss', reconstruction_loss)
|
|
self.log('identity_loss', identity_loss)
|
|
self.log('pose_loss', pose_loss)
|
|
self.log('gaze_loss', gaze_loss)
|
|
return generator_loss
|
|
|
|
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_embedding = calc_embedding(self.embedder, output_tensor, (0, 0, 0, 0))
|
|
validation = (nn.functional.cosine_similarity(source_embedding, output_embedding).mean() + 1) * 0.5
|
|
self.log('validation', validation)
|
|
return validation
|
|
|
|
def generate_preview(self, source_tensor : Tensor, target_tensor : Tensor, output_tensor : Tensor) -> None:
|
|
preview_limit = 8
|
|
preview_cells = []
|
|
|
|
for source_tensor, target_tensor, output_tensor in zip(source_tensor[:preview_limit], target_tensor[:preview_limit], output_tensor[:preview_limit]):
|
|
preview_cell = torch.cat([ source_tensor, target_tensor, output_tensor] , dim = 2)
|
|
preview_cells.append(preview_cell)
|
|
|
|
preview_cells = torch.cat(preview_cells, dim = 1).unsqueeze(0)
|
|
preview_grid = torchvision.utils.make_grid(preview_cells, normalize = True, scale_each = True)
|
|
self.logger.experiment.add_image('preview', preview_grid, self.global_step) # type:ignore[attr-defined]
|
|
|
|
|
|
def create_loaders(dataset : Dataset[Tensor]) -> Tuple[StatefulDataLoader[Tensor], StatefulDataLoader[Tensor]]:
|
|
batch_size = CONFIG.getint('training.loader', 'batch_size')
|
|
num_workers = CONFIG.getint('training.loader', 'num_workers')
|
|
|
|
training_dataset, validate_dataset = split_dataset(dataset)
|
|
training_loader = StatefulDataLoader(training_dataset, batch_size = batch_size, shuffle = True, num_workers = num_workers, drop_last = True, pin_memory = True, persistent_workers = True)
|
|
validation_loader = StatefulDataLoader(validate_dataset, batch_size = batch_size, shuffle = False, num_workers = num_workers, pin_memory = True, persistent_workers = True)
|
|
return training_loader, validation_loader
|
|
|
|
|
|
def split_dataset(dataset : Dataset[Tensor]) -> Tuple[Dataset[Tensor], Dataset[Tensor]]:
|
|
split_ratio = CONFIG.getfloat('training.loader', 'split_ratio')
|
|
dataset_size = len(dataset) # type:ignore[arg-type]
|
|
training_size = int(dataset_size * split_ratio)
|
|
validation_size = int(dataset_size - training_size)
|
|
training_dataset, validate_dataset = random_split(dataset, [ training_size, validation_size ])
|
|
return training_dataset, validate_dataset
|
|
|
|
|
|
def create_trainer() -> Trainer:
|
|
trainer_max_epochs = CONFIG.getint('training.trainer', 'max_epochs')
|
|
output_directory_path = CONFIG.get('training.output', 'directory_path')
|
|
output_file_pattern = CONFIG.get('training.output', 'file_pattern')
|
|
trainer_precision = CONFIG.get('training.trainer', 'precision')
|
|
logger = TensorBoardLogger('.logs', name = 'face_swapper')
|
|
|
|
return Trainer(
|
|
logger = logger,
|
|
log_every_n_steps = 10,
|
|
max_epochs = trainer_max_epochs,
|
|
precision = trainer_precision, # type:ignore[arg-type]
|
|
callbacks =
|
|
[
|
|
ModelCheckpoint(
|
|
monitor = 'generator_loss',
|
|
dirpath = output_directory_path,
|
|
filename = output_file_pattern,
|
|
every_n_train_steps = 1000,
|
|
save_top_k = 3,
|
|
save_last = True
|
|
)
|
|
],
|
|
val_check_interval = 1000
|
|
)
|
|
|
|
|
|
def train() -> None:
|
|
dataset_file_pattern = CONFIG.get('training.dataset', 'file_pattern')
|
|
dataset_batch_ratio = CONFIG.getfloat('training.dataset', 'batch_ratio')
|
|
output_resume_path = CONFIG.get('training.output', 'resume_path')
|
|
|
|
dataset = DynamicDataset(dataset_file_pattern, dataset_batch_ratio)
|
|
training_loader, validation_loader = create_loaders(dataset)
|
|
face_swapper_trainer = FaceSwapperTrainer()
|
|
trainer = create_trainer()
|
|
|
|
if os.path.isfile(output_resume_path):
|
|
trainer.fit(face_swapper_trainer, training_loader, validation_loader, ckpt_path = output_resume_path)
|
|
else:
|
|
trainer.fit(face_swapper_trainer, training_loader, validation_loader)
|