AI-based Surveillance System Project – Detecting Weapons in Live Video (Undergraduate Thesis)
Developed and implemented an AI-powered surveillance system to detect weapons in live video feeds as part of the undergraduate thesis. Utilized deep learning models such as YOLOv8 and Faster R-CNN for annotating and identifying objects within video frames. Integrated privacy-preserving features including automated face blurring and real-time alert mechanisms. • Labeled and annotated video data for weapon detection tasks • Applied bounding box and object detection techniques for accurate localization • Conducted continuous model evaluation using labeled datasets • Enhanced dataset diversity by including varied environmental video conditions