This project proposes an end-to-end framework for semi-supervised Anomaly Detection and Segmentation in images based on Deep Learning.

An AI-driven industrial anomaly detection system. Utilizes a fine-tuned YOLOv8 model trained on the MVTec AD dataset to inspect surfaces, sending real-time stop/start signals directly to Siemens S7 PLCs via Snap7.

Self-supervised learning for low-annotation surface defect detection and inspection