Open-source deep-learning library for visual anomaly detection: ready-to-train implementations of PatchCore, PaDiM, EfficientAD, FastFlow and more, with benchmarking, OpenVINO export and edge deployment — the standard toolkit for industrial defect detection.
Open-source deep-learning library for visual anomaly detection: ready-to-train implementations of PatchCore, PaDiM, EfficientAD, FastFlow and more, with benchmarking, OpenVINO export and edge deployment — the standard toolkit for industrial defect detection.
Save it, compare it, request a quote or record that you have used it — with an account.
Anomalib collects state-of-the-art unsupervised anomaly-detection algorithms — PatchCore, PaDiM, EfficientAD, FastFlow, Reverse Distillation, DRAEM and others — behind one consistent train/evaluate/deploy API. Born at Intel as part of the OpenVINO ecosystem (now hosted in the Open Edge Platform), it targets exactly the industrial QC scenario: train on good parts only, detect and localize defects at inference. It ships experiment management, benchmarking across MVTec AD and VisA datasets, hyperparameter optimization, and export to OpenVINO/ONNX/Torch for CPU-class edge inference. Version 2.x reworked the API on Lightning with improved metrics (per-image and per-pixel AUROC/PRO).
This product is part of the Birdwave Atlas community.
Contribute evaluations, share integrations and help improve compatibility data.
Discussions, evaluations, integrations and updates