Source code for mmagic.models.editors.dic.dic
# Copyright (c) OpenMMLab. All rights reserved.
from typing import Dict, List
import torch
from mmengine.optim import OptimWrapperDict
from mmagic.models.utils import set_requires_grad
from mmagic.registry import MODELS
from ..srgan import SRGAN
@MODELS.register_module()
[docs]class DIC(SRGAN):
"""DIC model for Face Super-Resolution.
Paper: Deep Face Super-Resolution with Iterative Collaboration between
Attentive Recovery and Landmark Estimation.
Args:
generator (dict): Config for the generator.
pixel_loss (dict): Config for the pixel loss.
align_loss (dict): Config for the align loss.
discriminator (dict): Config for the discriminator. Default: None.
gan_loss (dict): Config for the gan loss. Default: None.
feature_loss (dict): Config for the feature loss. Default: None.
train_cfg (dict): Config for train. Default: None.
test_cfg (dict): Config for testing. Default: None.
init_cfg (dict, optional): The weight initialized config for
:class:`BaseModule`. Default: None.
data_preprocessor (dict, optional): The pre-process config of
:class:`BaseDataPreprocessor`. Default: None.
"""
def __init__(self,
generator,
pixel_loss,
align_loss,
discriminator=None,
gan_loss=None,
feature_loss=None,
train_cfg=None,
test_cfg=None,
init_cfg=None,
data_preprocessor=None):
super().__init__(
generator=generator,
discriminator=discriminator,
gan_loss=gan_loss,
pixel_loss=pixel_loss,
train_cfg=train_cfg,
test_cfg=test_cfg,
init_cfg=init_cfg,
data_preprocessor=data_preprocessor)
self.align_loss = MODELS.build(align_loss)
self.feature_loss = MODELS.build(
feature_loss) if feature_loss else None
self.pixel_init = train_cfg.get('pixel_init', 0) if train_cfg else 0
[docs] def forward_tensor(self, inputs, data_samples=None, training=False):
"""Forward tensor. Returns result of simple forward.
Args:
inputs (torch.Tensor): batch input tensor collated by
:attr:`data_preprocessor`.
data_samples (List[BaseDataElement], optional):
data samples collated by :attr:`data_preprocessor`.
training (bool): Whether is training. Default: False.
Returns:
(Tensor | Tuple[List[Tensor]]): results of forward inference and
forward train.
"""
sr_list, heatmap_list = self.generator(inputs)
if training:
return sr_list, heatmap_list
else:
return sr_list[-1]
[docs] def if_run_d(self):
"""Calculates whether need to run the discriminator step."""
return self.step_counter >= self.pixel_init and super().if_run_d()
[docs] def g_step(self, batch_outputs, batch_gt_data):
"""G step of GAN: Calculate losses of generator.
Args:
batch_outputs (Tensor): Batch output of generator.
batch_gt_data (Tensor): Batch GT data.
Returns:
dict: Dict of losses.
"""
sr_list, heatmap_list = batch_outputs
gt, gt_heatmap = batch_gt_data
losses = dict()
# pix loss
for step, (sr, heatmap) in enumerate(zip(sr_list, heatmap_list)):
losses[f'loss_pixel_v{step}'] = self.pixel_loss(sr, gt)
losses[f'loss_align_v{step}'] = self.align_loss(
heatmap, gt_heatmap)
if self.step_counter >= self.pixel_init:
pred = sr_list[-1]
# perceptual loss
if self.feature_loss:
loss_feature = self.feature_loss(pred, gt)
losses['loss_feature'] = loss_feature
# gan loss for generator
if self.gan_loss and self.discriminator:
fake_g_pred = self.discriminator(pred)
losses['loss_gan'] = self.gan_loss(
fake_g_pred, target_is_real=True, is_disc=False)
return losses
[docs] def train_step(self, data: List[dict],
optim_wrapper: OptimWrapperDict) -> Dict[str, torch.Tensor]:
"""Train step of GAN-based method.
Args:
data (List[dict]): Data sampled from dataloader.
optim_wrapper (OptimWrapper): OptimWrapper instance
used to update model parameters.
Returns:
Dict[str, torch.Tensor]: A ``dict`` of tensor for logging.
"""
g_optim_wrapper = optim_wrapper['generator']
data = self.data_preprocessor(data, True)
batch_inputs = data['inputs']
data_samples = data['data_samples']
batch_gt_data = self.extract_gt_data(data_samples)
log_vars = dict()
with g_optim_wrapper.optim_context(self):
batch_outputs = self.forward_train(batch_inputs, data_samples)
if self.if_run_g():
set_requires_grad(self.discriminator, False)
log_vars_d = self.g_step_with_optim(
batch_outputs=batch_outputs,
batch_gt_data=batch_gt_data,
optim_wrapper=optim_wrapper)
log_vars.update(log_vars_d)
if self.if_run_d():
set_requires_grad(self.discriminator, True)
sr_list, _ = batch_outputs
gt, _ = batch_gt_data
for _ in range(self.disc_repeat):
# detach before function call to resolve PyTorch2.0 compile bug
log_vars_d = self.d_step_with_optim(
batch_outputs=sr_list[-1].detach(),
batch_gt_data=gt,
optim_wrapper=optim_wrapper)
log_vars.update(log_vars_d)
if 'loss' in log_vars:
log_vars.pop('loss')
self.step_counter += 1
return log_vars
@staticmethod
[docs] def extract_gt_data(data_samples):
"""extract gt data from data samples.
Args:
data_samples (list): List of DataSample.
Returns:
Tensor: Extract gt data.
"""
batch_gt_img = data_samples.gt_img
batch_gt_heatmap = data_samples.gt_heatmap
return [batch_gt_img, batch_gt_heatmap]