Skip to main content
GIS-Schools GIS-Schools
Anomaly & Defect Detection · GeoAI

Anomaly & Defect Detection

Highlight unusual visual patterns for inspection and review.

Anomaly & Defect Detection task illustration

What is anomaly & defect detection?

Anomaly detection compares observations with learned normal behavior. A high score indicates something unusual; it does not by itself identify a specific crack, cause or severity.

Approaches and starting model

Normal-only feature-memory methods such as PatchCore differ from supervised defect detectors trained with named damage classes. Select the approach according to available inspection labels.

Inputs and outputs

Consistently prepared inspection images enter the pipeline. Output includes an image-level score and a heatmap of unusual regions, not a guaranteed defect diagnosis or named defect class.

Where it helps in GIS and digital twins

Prioritize human review of component photos linked to asset IDs and inspection dates. Site-specific validation is required before applying this approach to cracks, corrosion or other infrastructure conditions.

How to start

Collect normal images covering allowed lighting, viewpoint and material variation. Fit the feature memory with Anomalib and calibrate thresholds on separate normal and defect examples. Preserve preprocessing and the fitted memory together.

How to evaluate results

Review missed defects and nuisance alarms by defect type and acquisition condition. Choose thresholds around the actual inspection workflow, retaining the original image beside every flag.

Limitations

A new background or illumination may look anomalous. A visually familiar but dangerous defect can be missed. Generic pretrained CNN weights alone do not constitute a fitted inspection model.

Official starting-model documentation

The category artwork illustrates its starting model; it is not an evaluated prediction.

Explore Anomaly & Defect Detection models