Real-time table occupancy tracking using YOLO11, ByteTrack, and live zone monitoring — built on a single camera feed.

Built a real-time computer vision system that monitors restaurant table occupancy using a single overhead camera — no sensors, no hardware changes, just a camera and intelligence.

The system detects customers the moment they sit down, tracks each table's occupancy state live, and maintains a real-time dashboard showing which tables are free, occupied, and how long each group has been seated — all processed directly on the camera feed at runtime.

Detects persons in real time using YOLO11 with per-detection confidence scoring

Maps each detection to its corresponding table zone using polygon ROI definitions

Tracks dwell time per table — counting seconds/minutes from the moment occupancy is detected

Classifies each table state live: Free (green) or Occupied (pink/red) with elapsed time

Displays a live sidebar dashboard showing all 7 tables, their states, total persons detected, occupancy count, and active alerts

Runs on a standard camera feed — no special hardware required

YOLO11 for real-time person detection with confidence thresholds

ByteTrack (via Ultralytics) for stable multi-object tracking across frames

Custom polygon zone definitions per table mapped to camera perspective

State machine logic per table: Free → Occupied → Alert

OpenCV for live frame annotation, zone overlays, and dashboard rendering

Python with Ultralytics, OpenCV, and custom business logic layer

A restaurant manager watching this feed knows instantly — without walking the floor — which tables need attention, which are free for new guests, and which groups have been waiting too long. The system replaces manual floor monitoring with a single intelligent camera.