Crowd-Density
python, opencv, yolov8, supervision
A real-time crowd monitoring system that detects density changes, bottlenecks, and reverse flow patterns from video.
view repositoryChallenge
Crowd risk builds quickly, so the system needed to surface meaningful movement signals fast enough to help with intervention.
Approach
I used object detection and tracking to monitor flow patterns across live footage, then structured the outputs around safety-relevant events instead of raw detections.
Outcome
The project translates dense scene analysis into actionable signals for crowd management in stadiums, events, and public spaces.
Highlights
- real-time detection pipeline
- bottleneck and reverse-flow monitoring
- built for event and public-space surveillance