Integrated Campus Safety and Security System
Real-time computer-vision system for person detection and role classification across campus security feeds.
Manually watching campus camera feeds does not scale, and useful events are easy to miss. The system automates detection so staff can respond instead of monitor.
A YOLOv8 model was trained for person detection and role classification, wrapped in optimized TensorFlow inference pipelines, and served through Django REST APIs. Accuracy and inference performance were tuned for production, and the models were deployed into a live environment alongside the rest of the security stack.
// AI & process
How AI accelerated this
A YOLOv8 model performs real-time person detection and role classification; TensorFlow inference pipelines are optimized for production latency and served to the Django backend via REST APIs.