Crowd Analytics Platform

Event Analytics System V1 — Privacy-safe edge AI for event analytics. Uses YOLO11 person detection with TensorRT optimization (2-3x faster), multi-object tracking via ByteTrack & BoT-SORT, line crossing counting with trajectory-based direction detection, RT-DETR crowd detection for zone-based density classification, and restricted area intrusion detection with real-time alerts.

This platform is designed for large-scale event monitoring, providing actionable insights while maintaining strict privacy standards. All processing happens at the edge with no face recognition — only anonymous tracking and counting. The system supports webcam, video files, and RTSP streams, making it versatile for various deployment scenarios.

Key Features

• YOLO11 + TensorRT real-time person detection
• ByteTrack & BoT-SORT multi-object tracking
• Line crossing counting with direction detection
• RT-DETR crowd density classification (GREEN/AMBER/RED zones)
• Restricted area intrusion alerts
• AWS IoT Core integration (MQTT)
• PM2 web dashboard for monitoring
• Privacy-first: no face recognition, anonymous tracking, edge processing only
• Supports webcam, video files, and RTSP streams

Demo Video

Project Gallery