SentinelAI School/College Security Module

Production-ready, modular, scalable, privacy-first AI security system for schools/colleges. Detects threats, triggers alerts, and provides a real-time dashboard.

Features

• RTSP ingest (camera feeds, or webcam for demo)
• Fight/violence detection (YOLOv8 + tracking)
• Gate accident, intrusion, abandoned object, fire/smoke, exam malpractice detection
• Alert pipeline & event engine
• Dashboard (Next.js + Tailwind)
• LLM summary & PDF export
• Docker deployable

Quick Start (Demo)

1. Copy .env.example to .env and fill in values as needed.
2. Install dependencies:
   • Python: pip install -r requirements.txt in backend, ai_worker, event_engine, alert_service
   • Dashboard: npm install in dashboard
3. Start backend: cd backend && uvicorn main:app --reload
4. Start AI worker: cd ai_worker && python worker.py worker fight (or any event type)
5. Start dashboard: cd dashboard && npm run dev
6. Open dashboard at http://localhost:3000
7. Test live camera feed at http://localhost:8000/live_feed

Event Types

• fight
• gate_accident
• intrusion
• abandoned_object
• fire_smoke
• exam_malpractice

Endpoints

• /event — POST event ingest
• /ws/alerts — WebSocket for real-time alerts
• /incident/{id}/summary — LLM summary
• /incident/{id}/pdf — PDF export
• /live_feed — MJPEG webcam stream

Scripts

• scripts/run_demo.sh — Start all services for demo (Linux/Mac)
• scripts/simulate_event.py — Send test event to backend

Folder Structure

• backend/ — FastAPI event API
• ai_worker/ — AI detection (YOLOv8, tracking, event logic)
• event_engine/ — Rule engine, scoring
• alert_service/ — Alert, evidence, LLM, S3
• dashboard/ — Next.js UI
• scripts/ — Demo/test scripts
• test_videos/ — Sample videos
• config/ — Config files

---

MIT License

SecVidAnalysis

SecVidAnalysis

SecVidAnalysis

CCTvAnalysis