Toggle Light / Dark / Auto color theme

Toggle table of contents sidebar

anomalib.utils.callbacks

Callbacks for Anomalib models.

Subpackages

• anomalib.utils.callbacks.nncf
  • anomalib.utils.callbacks.nncf.callback
  • anomalib.utils.callbacks.nncf.utils

Submodules

• anomalib.utils.callbacks.cdf_normalization
• anomalib.utils.callbacks.graph
• anomalib.utils.callbacks.metrics_configuration
• anomalib.utils.callbacks.min_max_normalization
• anomalib.utils.callbacks.model_loader
• anomalib.utils.callbacks.openvino
• anomalib.utils.callbacks.tiler_configuration
• anomalib.utils.callbacks.timer
• anomalib.utils.callbacks.visualizer_callback

Package Contents

Classes

CdfNormalizationCallback	Callback that standardizes the image-level and pixel-level anomaly scores.
MetricsConfigurationCallback	Metrics Configuration Callback.
MinMaxNormalizationCallback	Callback that normalizes the image-level and pixel-level anomaly scores using min-max normalization.
LoadModelCallback	Callback that loads the model weights from the state dict.
TilerConfigurationCallback	Tiler Configuration Callback.
TimerCallback	Callback that measures the training and testing time of a PyTorch Lightning module.
VisualizerCallback	Callback that visualizes the inference results of a model.

class anomalib.utils.callbacks.CdfNormalizationCallback

Bases: pytorch_lightning.Callback

Callback that standardizes the image-level and pixel-level anomaly scores.

on_test_start(self, _trainer: pytorch_lightning.Trainer, pl_module: anomalib.models.components.AnomalyModule) → None

Called when the test begins.

on_validation_epoch_start(self, trainer: pytorch_lightning.Trainer, pl_module: anomalib.models.components.AnomalyModule) → None

Called when the validation starts after training.

Use the current model to compute the anomaly score distributions of the normal training data. This is needed after every epoch, because the statistics must be stored in the state dict of the checkpoint file.

on_validation_batch_end(self, _trainer: pytorch_lightning.Trainer, pl_module: anomalib.models.components.AnomalyModule, outputs: Optional[pytorch_lightning.utilities.types.STEP_OUTPUT], _batch: Any, _batch_idx: int, _dataloader_idx: int) → None

Called when the validation batch ends, standardizes the predicted scores and anomaly maps.

on_test_batch_end(self, _trainer: pytorch_lightning.Trainer, pl_module: anomalib.models.components.AnomalyModule, outputs: Optional[pytorch_lightning.utilities.types.STEP_OUTPUT], _batch: Any, _batch_idx: int, _dataloader_idx: int) → None

Called when the test batch ends, normalizes the predicted scores and anomaly maps.

on_predict_batch_end(self, _trainer: pytorch_lightning.Trainer, pl_module: anomalib.models.components.AnomalyModule, outputs: Dict, _batch: Any, _batch_idx: int, _dataloader_idx: int) → None

Called when the predict batch ends, normalizes the predicted scores and anomaly maps.

_collect_stats(self, trainer, pl_module)

Collect the statistics of the normal training data.

Create a trainer and use it to predict the anomaly maps and scores of the normal training data. Then

estimate the distribution of anomaly scores for normal data at the image and pixel level by computing the mean and standard deviations. A dictionary containing the computed statistics is stored in self.stats.

static _create_inference_model(pl_module)

Create a duplicate of the PL module that can be used to perform inference on the training set.

static _standardize_batch(outputs: pytorch_lightning.utilities.types.STEP_OUTPUT, pl_module) → None

static _normalize_batch(outputs: pytorch_lightning.utilities.types.STEP_OUTPUT, pl_module: anomalib.models.components.AnomalyModule) → None

class anomalib.utils.callbacks.MetricsConfigurationCallback(adaptive_threshold: bool, default_image_threshold: Optional[float] = None, default_pixel_threshold: Optional[float] = None, image_metric_names: Optional[List[str]] = None, pixel_metric_names: Optional[List[str]] = None, normalization_method: str = 'min_max')

Bases: pytorch_lightning.callbacks.Callback

Metrics Configuration Callback.

setup(self, _trainer: pytorch_lightning.Trainer, pl_module: pytorch_lightning.LightningModule, stage: Optional[str] = None) → None

Setup image and pixel-level AnomalibMetricsCollection within Anomalib Model.

Parameters

• _trainer (pl.Trainer) – PyTorch Lightning Trainer

• pl_module (pl.LightningModule) – Anomalib Model that inherits pl LightningModule.

• stage (Optional[str], optional) – fit, validate, test or predict. Defaults to None.

class anomalib.utils.callbacks.MinMaxNormalizationCallback

Bases: pytorch_lightning.Callback

Callback that normalizes the image-level and pixel-level anomaly scores using min-max normalization.

on_test_start(self, _trainer: pytorch_lightning.Trainer, pl_module: anomalib.models.components.AnomalyModule) → None

Called when the test begins.

on_validation_batch_end(self, _trainer: pytorch_lightning.Trainer, pl_module: anomalib.models.components.AnomalyModule, outputs: pytorch_lightning.utilities.types.STEP_OUTPUT, _batch: Any, _batch_idx: int, _dataloader_idx: int) → None

Called when the validation batch ends, update the min and max observed values.

on_test_batch_end(self, _trainer: pytorch_lightning.Trainer, pl_module: anomalib.models.components.AnomalyModule, outputs: pytorch_lightning.utilities.types.STEP_OUTPUT, _batch: Any, _batch_idx: int, _dataloader_idx: int) → None

Called when the test batch ends, normalizes the predicted scores and anomaly maps.

on_predict_batch_end(self, _trainer: pytorch_lightning.Trainer, pl_module: anomalib.models.components.AnomalyModule, outputs: Dict, _batch: Any, _batch_idx: int, _dataloader_idx: int) → None

Called when the predict batch ends, normalizes the predicted scores and anomaly maps.

static _normalize_batch(outputs, pl_module)

Normalize a batch of predictions.

class anomalib.utils.callbacks.LoadModelCallback(weights_path)

Bases: pytorch_lightning.Callback

Callback that loads the model weights from the state dict.

on_test_start(self, _trainer, pl_module: anomalib.models.components.AnomalyModule) → None

Call when the test begins.

Loads the model weights from weights_path into the PyTorch module.

on_predict_start(self, _trainer, pl_module: anomalib.models.components.AnomalyModule) → None

Call when inference begins.

Loads the model weights from weights_path into the PyTorch module.

class anomalib.utils.callbacks.TilerConfigurationCallback(enable: bool = False, tile_size: Union[int, Sequence] = 256, stride: Optional[Union[int, Sequence]] = None, remove_border_count: int = 0, mode: str = 'padding', tile_count: int = 4)

Bases: pytorch_lightning.callbacks.Callback

Tiler Configuration Callback.

setup(self, _trainer: pytorch_lightning.Trainer, pl_module: pytorch_lightning.LightningModule, stage: Optional[str] = None) → None

Setup Tiler object within Anomalib Model.

Parameters

• _trainer (pl.Trainer) – PyTorch Lightning Trainer

• pl_module (pl.LightningModule) – Anomalib Model that inherits pl LightningModule.

• stage (Optional[str], optional) – fit, validate, test or predict. Defaults to None.

Raises

ValueError – When Anomalib Model doesn’t contain Tiler object, it means the model doesn not support tiling operation.

class anomalib.utils.callbacks.TimerCallback

Bases: pytorch_lightning.Callback

Callback that measures the training and testing time of a PyTorch Lightning module.

on_fit_start(self, trainer: pytorch_lightning.Trainer, pl_module: pytorch_lightning.LightningModule) → None

Call when fit begins.

Sets the start time to the time training started.

Parameters

• trainer (Trainer) – PyTorch Lightning trainer.

• pl_module (LightningModule) – Current training module.

Returns

None

on_fit_end(self, trainer: pytorch_lightning.Trainer, pl_module: pytorch_lightning.LightningModule) → None

Call when fit ends.

Prints the time taken for training.

Parameters

• trainer (Trainer) – PyTorch Lightning trainer.

• pl_module (LightningModule) – Current training module.

Returns

None

on_test_start(self, trainer: pytorch_lightning.Trainer, pl_module: pytorch_lightning.LightningModule) → None

Call when the test begins.

Sets the start time to the time testing started. Goes over all the test dataloaders and adds the number of images in each.

Parameters

• trainer (Trainer) – PyTorch Lightning trainer.

• pl_module (LightningModule) – Current training module.

Returns

None

on_test_end(self, trainer: pytorch_lightning.Trainer, pl_module: pytorch_lightning.LightningModule) → None

Call when the test ends.

Prints the time taken for testing and the throughput in frames per second.

Parameters

• trainer (Trainer) – PyTorch Lightning trainer.

• pl_module (LightningModule) – Current training module.

Returns

None

class anomalib.utils.callbacks.VisualizerCallback(task: str, mode: str, image_save_path: str, inputs_are_normalized: bool = True, show_images: bool = False, log_images: bool = True, save_images: bool = True)

Bases: pytorch_lightning.Callback

Callback that visualizes the inference results of a model.

The callback generates a figure showing the original image, the ground truth segmentation mask, the predicted error heat map, and the predicted segmentation mask.

To save the images to the filesystem, add the ‘local’ keyword to the project.log_images_to parameter in the config.yaml file.

_add_to_logger(self, image: numpy.ndarray, module: anomalib.models.components.AnomalyModule, trainer: pytorch_lightning.Trainer, filename: pathlib.Path)

Log image from a visualizer to each of the available loggers in the project.

Parameters

• image (np.ndarray) – Image that should be added to the loggers.

• module (AnomalyModule) – Anomaly module.

• trainer (Trainer) – Pytorch Lightning trainer which holds reference to logger

• filename (Path) – Path of the input image. This name is used as name for the generated image.

on_predict_batch_end(self, _trainer: pytorch_lightning.Trainer, _pl_module: anomalib.models.components.AnomalyModule, outputs: Optional[pytorch_lightning.utilities.types.STEP_OUTPUT], _batch: Any, _batch_idx: int, _dataloader_idx: int) → None

Show images at the end of every batch.

Parameters

• _trainer (Trainer) – Pytorch lightning trainer object (unused).

• _pl_module (LightningModule) – Lightning modules derived from BaseAnomalyLightning object as

• images. (currently only they support logging) –

• outputs (Dict[str, Any]) – Outputs of the current test step.

• _batch (Any) – Input batch of the current test step (unused).

• _batch_idx (int) – Index of the current test batch (unused).

• _dataloader_idx (int) – Index of the dataloader that yielded the current batch (unused).

on_test_batch_end(self, trainer: pytorch_lightning.Trainer, pl_module: anomalib.models.components.AnomalyModule, outputs: Optional[pytorch_lightning.utilities.types.STEP_OUTPUT], _batch: Any, _batch_idx: int, _dataloader_idx: int) → None

Log images at the end of every batch.

Parameters

• trainer (Trainer) – Pytorch lightning trainer object (unused).

• pl_module (LightningModule) – Lightning modules derived from BaseAnomalyLightning object as

• images. (currently only they support logging) –

• outputs (Dict[str, Any]) – Outputs of the current test step.

• _batch (Any) – Input batch of the current test step (unused).

• _batch_idx (int) – Index of the current test batch (unused).

• _dataloader_idx (int) – Index of the dataloader that yielded the current batch (unused).

on_test_end(self, _trainer: pytorch_lightning.Trainer, pl_module: anomalib.models.components.AnomalyModule) → None

Sync logs.

Currently only AnomalibWandbLogger is called from this method. This is because logging as a single batch ensures that all images appear as part of the same step.

Parameters

• _trainer (pl.Trainer) – Pytorch Lightning trainer (unused)

• pl_module (AnomalyModule) – Anomaly module