FlowSync AI is a real-time crowd intelligence system designed for large-scale venues like stadiums, concerts, and public events.
It enables density-aware routing, queue prediction, and operational decision-making using live or simulated crowd dataโhelping organizers manage 100,000+ attendees efficiently in real time.
Large events face:
- Unpredictable crowd congestion
- Long queues at food/restrooms
- Unsafe bottlenecks during peak moments
- Lack of real-time visibility for operators
Most systems today are reactive, not predictive.
FlowSync AI is predictive + adaptive.
It continuously:
- Monitors crowd density
- Predicts congestion hotspots
- Suggests optimal routes
- Recommends operator interventions
- Zone-based density heatmaps
- Live crowd simulation / ingestion
- Hotspot detection
- Density-weighted pathfinding
- Avoids congested zones dynamically
- Phase-aware navigation (entry / halftime / exit)
- Predict wait times for stalls/restrooms
- Suggest faster alternatives
- High-risk zones detection
- Actionable interventions
- Real-time alerts system
- Surge simulation
- Food rush spike
- Emergency evacuation mode
- AI optimization mode
Tested using Narendra Modi Stadium (Ahmedabad) layout:
- Real geographic structure
- Gate-based entry/exit logic
- Zone-mapped crowd simulation
- Narendra Modi Stadium
- Wembley Stadium
- MetLife Stadium
- Melbourne Cricket Ground
- Camp Nou
- Times Square
- Burj Khalifa
- Eiffel Tower
- Tirupati Temple
- Kumbh Mela Grounds
| Layer | Tech |
|---|---|
| Frontend | React + TypeScript + Vite |
| Backend | FastAPI (Python) |
| Deployment | Google Cloud Run |
| Maps | Google Maps Platform |
| Algorithms | Grid-based pathfinding + density cost weighting |
| Data Mode | Hybrid (Simulated + Live Ingest API) |
- ๐ Frontend: https://flowsync-frontend-796656775802.asia-south1.run.app
- ๐ Backend API: https://flowsync-backend-796656775802.asia-south1.run.app
- ๐ API Docs: https://flowsync-backend-796656775802.asia-south1.run.app/docs
Frontend (React) -> Backend API (FastAPI) -> Crowd Simulator / Live Data Ingestion -> Routing + Prediction Engine
cd backend
pip install -r requirements.txt
uvicorn app.main:app --reload --port 8000cd frontend
npm install
npm run dev- Frontend โ http://localhost:5173
- Backend Docs โ http://localhost:8000/docs
| Endpoint | Description |
|---|---|
/snapshot |
Current crowd + queue state |
/heatmap |
Zone density map |
/journey |
Smart route planning |
/venue-map |
Venue structure |
/staff-actions |
AI operator suggestions |
/alerts |
Real-time alerts |
/demo-control |
Trigger simulation scenarios |
/ingest/live-snapshot |
Push real telemetry |
/data-source |
Live vs simulated data status |
{
"generated_at": "2026-04-15T12:30:00Z",
"zones": [
{ "zone_id": "zone-1", "density_score": 88 }
],
"queues": [
{ "stall_id": "stall-1", "wait_time_minutes": 12 }
]
}Trigger via /demo-control:
surge-zone-1food-rushemergency-modeoptimize-crowd
- Real-world venue simulation (not dummy/demo-only data)
- AI-driven routing and prediction logic
- Full-stack deployment on Google Cloud Run
- Covers both attendee experience + operator decision-making
- Designed to scale with real telemetry (IoT, CCTV, WiFi tracking)
- API keys (Google Maps) are restricted by domain
- Backend CORS is configured for the deployed frontend
- System works in simulated mode by default if no live data is ingested
MIT License