Post-processing

Post-processing#

Base Post-processor

Base class for post-processing.

Learn more »

Base Post-processor
One-class Post-processor

Post-processor for one-class anomaly detection.

Learn more »

One-class Post-processor

Base Post-processor#

Base class for post-processing anomaly detection results.

This module provides the abstract base class PostProcessor that defines the interface for post-processing anomaly detection outputs.

The post-processors handle:

  • Normalizing anomaly scores

  • Thresholding and anomaly classification

  • Mask generation and refinement

  • Result aggregation and formatting

Example

>>> from anomalib.post_processing import PostProcessor
>>> class MyPostProcessor(PostProcessor):
...     def forward(self, batch):
...         # Post-process the batch
...         return batch

The post-processors are implemented as both torch.nn.Module and lightning.pytorch.Callback to support both inference and training workflows.

class anomalib.post_processing.base.PostProcessor(*args, **kwargs)#

Bases: Module, Callback, ABC

Base class for post-processing anomaly detection results.

The post-processor is implemented as both a torch.nn.Module and lightning.pytorch.Callback to support inference and training workflows. It handles tasks like score normalization, thresholding, and mask refinement.

The class must be inherited and the forward() method must be implemented to define the post-processing logic.

Example

>>> from anomalib.post_processing import PostProcessor
>>> class MyPostProcessor(PostProcessor):
...     def forward(self, batch):
...         # Normalize scores between 0 and 1
...         batch.anomaly_scores = normalize(batch.anomaly_scores)
...         return batch
abstract forward(batch)#

Post-process a batch of model predictions.

Parameters:

batch (anomalib.data.InferenceBatch) – Batch containing model predictions and metadata.

Returns:

Post-processed batch with

normalized scores, thresholded predictions, and/or refined masks.

Return type:

anomalib.data.InferenceBatch

Raises:

NotImplementedError – This is an abstract method that must be implemented by subclasses.

One-class Post-processor#

Post-processing module for one-class anomaly detection results.

This module provides post-processing functionality for one-class anomaly detection outputs through the OneClassPostProcessor class.

The post-processor handles:

  • Normalizing image and pixel-level anomaly scores

  • Computing adaptive thresholds for anomaly classification

  • Applying sensitivity adjustments to thresholds

  • Formatting results for downstream use

Example

>>> from anomalib.post_processing import OneClassPostProcessor
>>> post_processor = OneClassPostProcessor(image_sensitivity=0.5)
>>> predictions = post_processor(anomaly_maps=anomaly_maps)
class anomalib.post_processing.one_class.OneClassPostProcessor(image_sensitivity=None, pixel_sensitivity=None, **kwargs)#

Bases: PostProcessor

Post-processor for one-class anomaly detection.

This class handles post-processing of anomaly detection results by:

  • Normalizing image and pixel-level anomaly scores

  • Computing adaptive thresholds for anomaly classification

  • Applying sensitivity adjustments to thresholds

  • Formatting results for downstream use

Parameters:

  • image_sensitivity (float | None, optional) – Sensitivity value for image-level predictions. Higher values make the model more sensitive to anomalies. Defaults to None.

  • pixel_sensitivity (float | None, optional) – Sensitivity value for pixel-level predictions. Higher values make the model more sensitive to anomalies. Defaults to None.

  • **kwargs – Additional keyword arguments passed to parent class.

Example

>>> from anomalib.post_processing import OneClassPostProcessor
>>> post_processor = OneClassPostProcessor(image_sensitivity=0.5)
>>> predictions = post_processor(anomaly_maps=anomaly_maps)
forward(predictions)#

Post-process model predictions.

Parameters:

predictions (InferenceBatch) – Batch containing model predictions.

Returns:

Post-processed batch with normalized scores and

thresholded predictions.

Return type:

InferenceBatch

Raises:

ValueError – If neither pred_score nor anomaly_map is provided.

property image_max: float#

Get the maximum value for image-level normalization.

Returns:

Maximum image-level value.

Return type:

float

property image_min: float#

Get the minimum value for image-level normalization.

Returns:

Minimum image-level value.

Return type:

float

normalize_batch(batch)#

Normalize predicted scores and anomaly maps.

Parameters:

batch (Batch) – Batch containing model predictions.

Return type:

None

property normalized_image_threshold: float#

Get the normalized image-level threshold.

Returns:

Normalized image-level threshold value, adjusted by sensitivity.

Return type:

float

property normalized_pixel_threshold: float#

Get the normalized pixel-level threshold.

Returns:

Normalized pixel-level threshold value, adjusted by sensitivity.

Return type:

float

on_predict_batch_end(trainer, pl_module, outputs, *args, **kwargs)#

Normalize predicted scores and anomaly maps.

Parameters:

  • trainer (Trainer) – PyTorch Lightning trainer instance.

  • pl_module (LightningModule) – PyTorch Lightning module instance.

  • outputs (Batch) – Batch containing model predictions.

  • *args – Variable length argument list.

  • **kwargs – Arbitrary keyword arguments.

Return type:

None

on_test_batch_end(trainer, pl_module, outputs, *args, **kwargs)#

Apply post-processing steps to current batch of predictions.

Parameters:

  • trainer (Trainer) – PyTorch Lightning trainer instance.

  • pl_module (LightningModule) – PyTorch Lightning module instance.

  • outputs (Batch) – Batch containing model predictions.

  • *args – Variable length argument list.

  • **kwargs – Arbitrary keyword arguments.

Return type:

None

on_validation_batch_end(trainer, pl_module, outputs, *args, **kwargs)#

Update normalization and thresholding metrics using batch output.

Parameters:

  • trainer (Trainer) – PyTorch Lightning trainer instance.

  • pl_module (LightningModule) – PyTorch Lightning module instance.

  • outputs (Batch) – Batch containing model predictions and ground truth.

  • *args – Variable length argument list.

  • **kwargs – Arbitrary keyword arguments.

Return type:

None

on_validation_epoch_end(trainer, pl_module)#

Compute final threshold and normalization values.

Parameters:

  • trainer (Trainer) – PyTorch Lightning trainer instance.

  • pl_module (LightningModule) – PyTorch Lightning module instance.

Return type:

None

property pixel_max: float#

Get the maximum value for pixel-level normalization.

Returns:

Maximum pixel-level value.

Return type:

float

property pixel_min: float#

Get the minimum value for pixel-level normalization.

Returns:

Minimum pixel-level value.

Return type:

float

post_process_batch(batch)#

Post-process a batch of predictions.

Applies normalization and thresholding to the batch predictions.

Parameters:

batch (Batch) – Batch containing model predictions.

Return type:

None

property raw_image_threshold: float#

Get the raw image-level threshold.

Returns:

Raw image-level threshold value.

Return type:

float

property raw_pixel_threshold: float#

Get the raw pixel-level threshold.

Returns:

Raw pixel-level threshold value.

Return type:

float

threshold_batch(batch)#

Apply thresholding to batch predictions.

Parameters:

batch (Batch) – Batch containing model predictions.

Return type:

None