amazon-science/patchcore-inspection

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1373 stars · 252 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Official implementation of PatchCore, a deep-learning method for industrial image anomaly detection and localization from Roth et al. (2021). It trains and evaluates models on datasets like MVTec AD, achieving state-of-the-art image-level and pixel-level anomaly detection AUROC.

Use cases

• detect defects in industrial product images
• segment anomalous regions in manufacturing inspection photos
• train an anomaly detection model on MVTec AD
• evaluate pretrained PatchCore models on defect datasets
• replicate PatchCore paper results
• build visual quality-control inspection pipelines

When to choose

• you need high-accuracy industrial anomaly detection or defect localization on images
• you want to reproduce or build on the PatchCore paper
• you have a GPU and a labeled-normal image dataset like MVTec AD

When to avoid

• you need a production-ready maintained product rather than research code
• your task is not image-based anomaly detection
• you cannot use GPU acceleration or PyTorch

Facets

library · maturity maintenance

machine-learning computer-vision image-processing deep-learning computer-vision machine-learning deep-learning image-processing python anomaly-detection industrial-inspection patchcore defect-detection research-code mvtec-ad faiss linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="amazon-science/patchcore-inspection")