M-3LAB/awesome-industrial-anomaly-detection

Curates a comprehensive taxonomy of industrial anomaly detection methods—including feature-embedding, reconstruction-based, and supervised approaches—organized by architecture type (teacher-student, one-class classification, diffusion models, etc.). Covers emerging research directions like zero-shot detection, 3D anomaly detection, anomaly synthesis benchmarks, and multimodal large language model evaluation for quality inspection. Maintains linked benchmarks (IM-IAD) and code implementations alongside peer-reviewed papers from major venues (CVPR, ICCV, NeurIPS).

3,365 stars. Actively maintained with 8 commits in the last 30 days.