Overview
Restaurant operators face constant challenges managing seating capacity and customer flow, often relying on staff manually tracking available tables. CaféVision automates this entirely — a ceiling-mounted camera feed is processed by YOLOv11x in real time to detect occupied and unoccupied tables throughout the dining space.
The system integrates detection outputs into reporting tools providing actionable insights for staffing, reservation management, and layout optimization.
Technical Pipeline
01
Video Ingestion
Live camera feed captured frame-by-frame using OpenCV pipeline
02
YOLOv11x Detection
Person and empty table regions detected simultaneously per frame
03
Occupancy Logic
Rule-based inference from bounding box overlap with predefined table zones
04
Adaptive Training
Data augmentation for occlusion, lighting variation, and viewing angle robustness
05
Dashboard Output
Real-time occupancy counts fed to reporting layer for operational decisions
Live Detection Demos
Real-time YOLOv11x detection running on actual restaurant video feeds. Click to watch on YouTube.
Impact & Applications
- Customer Flow Optimization — real-time seat availability reduces wait times and improves turnover
- Staffing Intelligence — occupancy trends inform shift scheduling and resource allocation
- Contactless Monitoring — purely camera-based passive sensing, no wearables or tags required
- Scalability — architecture supports multi-camera deployments across large restaurant floors