Toggle Light / Dark / Auto color theme

Toggle table of contents sidebar

Source code for anomalib.models.ganomaly.lightning_model

"""GANomaly: Semi-Supervised Anomaly Detection via Adversarial Training.

https://arxiv.org/abs/1805.06725
"""

# Copyright (C) 2020 Intel Corporation
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions
# and limitations under the License.

import logging
from typing import Dict, List, Tuple, Union

import torch
from omegaconf import DictConfig, ListConfig
from pytorch_lightning.callbacks import EarlyStopping
from pytorch_lightning.utilities.cli import MODEL_REGISTRY
from torch import Tensor, optim

from anomalib.models.components import AnomalyModule
from anomalib.models.ganomaly.loss import DiscriminatorLoss, GeneratorLoss

from .torch_model import GanomalyModel

logger = logging.getLogger(__name__)


@MODEL_REGISTRY
class Ganomaly(AnomalyModule):
    """PL Lightning Module for the GANomaly Algorithm.

    Args:
        batch_size (int): Batch size.
        input_size (Tuple[int,int]): Input dimension.
        n_features (int): Number of features layers in the CNNs.
        latent_vec_size (int): Size of autoencoder latent vector.
        extra_layers (int, optional): Number of extra layers for encoder/decoder. Defaults to 0.
        add_final_conv_layer (bool, optional): Add convolution layer at the end. Defaults to True.
        wadv (int, optional): Weight for adversarial loss. Defaults to 1.
        wcon (int, optional): Image regeneration weight. Defaults to 50.
        wenc (int, optional): Latent vector encoder weight. Defaults to 1.
    """

    def __init__(
        self,
        batch_size: int,
        input_size: Tuple[int, int],
        n_features: int,
        latent_vec_size: int,
        extra_layers: int = 0,
        add_final_conv_layer: bool = True,
        wadv: int = 1,
        wcon: int = 50,
        wenc: int = 1,
        lr: float = 0.0002,
        beta1: float = 0.5,
        beta2: float = 0.999,
    ):
        super().__init__()

        self.model: GanomalyModel = GanomalyModel(
            input_size=input_size,
            num_input_channels=3,
            n_features=n_features,
            latent_vec_size=latent_vec_size,
            extra_layers=extra_layers,
            add_final_conv_layer=add_final_conv_layer,
        )

        self.real_label = torch.ones(size=(batch_size,), dtype=torch.float32)
        self.fake_label = torch.zeros(size=(batch_size,), dtype=torch.float32)

        self.min_scores: Tensor = torch.tensor(float("inf"), dtype=torch.float32)  # pylint: disable=not-callable
        self.max_scores: Tensor = torch.tensor(float("-inf"), dtype=torch.float32)  # pylint: disable=not-callable

        self.generator_loss = GeneratorLoss(wadv, wcon, wenc)
        self.discriminator_loss = DiscriminatorLoss()

        # TODO: LR should be part of optimizer in config.yaml! Since ganomaly has custom
        #   optimizer this is to be addressed later.
        self.learning_rate = lr
        self.beta1 = beta1
        self.beta2 = beta2

    def _reset_min_max(self):
        """Resets min_max scores."""
        self.min_scores = torch.tensor(float("inf"), dtype=torch.float32)  # pylint: disable=not-callable
        self.max_scores = torch.tensor(float("-inf"), dtype=torch.float32)  # pylint: disable=not-callable

    def configure_optimizers(self) -> List[optim.Optimizer]:
        """Configures optimizers for each decoder.

        Note:
            This method is used for the existing CLI.
            When PL CLI is introduced, configure optimizers method will be
                deprecated, and optimizers will be configured from either
                config.yaml file or from CLI.

        Returns:
            Optimizer: Adam optimizer for each decoder
        """
        optimizer_d = optim.Adam(
            self.model.discriminator.parameters(),
            lr=self.learning_rate,
            betas=(self.beta1, self.beta2),
        )
        optimizer_g = optim.Adam(
            self.model.generator.parameters(),
            lr=self.learning_rate,
            betas=(self.beta1, self.beta2),
        )
        return [optimizer_d, optimizer_g]

    def training_step(self, batch, _, optimizer_idx):  # pylint: disable=arguments-differ
        """Training step.

        Args:
            batch (Dict): Input batch containing images.
            optimizer_idx (int): Optimizer which is being called for current training step.

        Returns:
            Dict[str, Tensor]: Loss
        """
        # forward pass
        padded, fake, latent_i, latent_o = self.model(batch["image"])
        pred_real, _ = self.model.discriminator(padded)

        if optimizer_idx == 0:  # Discriminator
            pred_fake, _ = self.model.discriminator(fake.detach())
            loss = self.discriminator_loss(pred_real, pred_fake)
        else:  # Generator
            pred_fake, _ = self.model.discriminator(fake)
            loss = self.generator_loss(latent_i, latent_o, padded, fake, pred_real, pred_fake)

        return {"loss": loss}

    def on_validation_start(self) -> None:
        """Reset min and max values for current validation epoch."""
        self._reset_min_max()
        return super().on_validation_start()

    def validation_step(self, batch, _) -> Dict[str, Tensor]:  # type: ignore # pylint: disable=arguments-differ
        """Update min and max scores from the current step.

        Args:
            batch (Dict[str, Tensor]): Predicted difference between z and z_hat.

        Returns:
            Dict[str, Tensor]: batch
        """
        batch["pred_scores"] = self.model(batch["image"])
        self.max_scores = max(self.max_scores, torch.max(batch["pred_scores"]))
        self.min_scores = min(self.min_scores, torch.min(batch["pred_scores"]))
        return batch

    def validation_epoch_end(self, outputs):
        """Normalize outputs based on min/max values."""
        logger.info("Normalizing validation outputs based on min/max values.")
        for prediction in outputs:
            prediction["pred_scores"] = self._normalize(prediction["pred_scores"])
        super().validation_epoch_end(outputs)
        return outputs

    def on_test_start(self) -> None:
        """Reset min max values before test batch starts."""
        self._reset_min_max()
        return super().on_test_start()

    def test_step(self, batch, _):
        """Update min and max scores from the current step."""
        super().test_step(batch, _)
        self.max_scores = max(self.max_scores, torch.max(batch["pred_scores"]))
        self.min_scores = min(self.min_scores, torch.min(batch["pred_scores"]))
        return batch

    def test_epoch_end(self, outputs):
        """Normalize outputs based on min/max values."""
        logger.info("Normalizing test outputs based on min/max values.")
        for prediction in outputs:
            prediction["pred_scores"] = self._normalize(prediction["pred_scores"])
        super().test_epoch_end(outputs)
        return outputs

    def _normalize(self, scores: Tensor) -> Tensor:
        """Normalize the scores based on min/max of entire dataset.

        Args:
            scores (Tensor): Un-normalized scores.

        Returns:
            Tensor: Normalized scores.
        """
        scores = (scores - self.min_scores.to(scores.device)) / (
            self.max_scores.to(scores.device) - self.min_scores.to(scores.device)
        )
        return scores


class GanomalyLightning(Ganomaly):
    """PL Lightning Module for the GANomaly Algorithm.

    Args:
        hparams (Union[DictConfig, ListConfig]): Model params
    """

    def __init__(self, hparams: Union[DictConfig, ListConfig]) -> None:

        super().__init__(
            batch_size=hparams.dataset.train_batch_size,
            input_size=hparams.model.input_size,
            n_features=hparams.model.n_features,
            latent_vec_size=hparams.model.latent_vec_size,
            extra_layers=hparams.model.extra_layers,
            add_final_conv_layer=hparams.model.add_final_conv,
            wadv=hparams.model.wadv,
            wcon=hparams.model.wcon,
            wenc=hparams.model.wenc,
            lr=hparams.model.lr,
            beta1=hparams.model.beta1,
            beta2=hparams.model.beta2,
        )
        self.hparams: Union[DictConfig, ListConfig]  # type: ignore
        self.save_hyperparameters(hparams)

    def configure_callbacks(self):
        """Configure model-specific callbacks.

        Note:
            This method is used for the existing CLI.
            When PL CLI is introduced, configure callback method will be
                deprecated, and callbacks will be configured from either
                config.yaml file or from CLI.
        """
        early_stopping = EarlyStopping(
            monitor=self.hparams.model.early_stopping.metric,
            patience=self.hparams.model.early_stopping.patience,
            mode=self.hparams.model.early_stopping.mode,
        )
        return [early_stopping]