Diffusion-based Industrial Anomaly Detection

This project centers on industrial anomaly detection for real-world inspection settings. I explore how diffusion models, normality prior learning, semantic conditioning, and segmentation mechanisms can be combined to improve anomaly detection and defect localization.