Built for CSI Core Projects Initiative 2026 Domain: Cybersecurity + AI + Full Stack Duration: 25 Days | Team Size: 4
SentinAI converts passive CCTV cameras into active, tamper-resistant safety systems that detect threats in real time and raise alerts even during network or camera attacks.
-
Traditional CCTV systems only record incidents; they don’t prevent them.
-
Many cameras are easily disabled via:
- Feed looping
- Black screen attacks
- Network cable cuts
-
This leads to delayed response, lost evidence, and unsafe environments.
Cities today have thousands of cameras that witness tragedies but do nothing to stop them.
SentinAI is an edge-based smart surveillance system that:
- Detects fire, accidents, and SOS gestures using AI
- Detects camera tampering and feed manipulation
- Triggers local alarms instantly, even if the internet is cut
- Displays alerts and logs on a simple live dashboard
The system works independently of the network, making it reliable in real-world attack scenarios.
SentinAI focuses on securing surveillance infrastructure, not just AI detection.
It ensures:
- Availability → Works even during network failure
- Integrity → Detects feed manipulation and looping
- Resilience → Edge-based processing prevents single-point failure
This places the project under Cyber-Physical Security & Surveillance Security, a modern cybersecurity domain.
- Fire & accident detection (YOLO – pre-trained)
- SOS / distress gesture detection (MediaPipe)
- Black screen detection
- Looped / frozen feed detection
- Camera heartbeat monitoring
- Local buzzer / siren (works without internet)
- Dashboard alerts
- Time-stamped incident logging
- Live camera feed
- Real-time alerts
- Event history log
- Camera captures live video
- Edge device processes feed locally
- AI detects anomalies (fire / accident / SOS)
- Security module checks feed integrity
- Alerts triggered instantly
- Events logged and shown on dashboard
| Layer | Technology |
|---|---|
| Edge Device | Raspberry Pi / Jetson Nano |
| AI Models | YOLO, MediaPipe |
| CV Processing | OpenCV |
| Backend | Python, Flask / FastAPI |
| Frontend | React.js / Flutter |
| Storage | SQLite / Local Storage |
| Alerts | Buzzer, Notifications |
| Member | Responsibility |
|---|---|
| Anusha | AI & Computer Vision |
| Shravya | Cybersecurity & Edge Security |
| Sujesh | Backend Development |
| Bhuvan | Frontend & Dashboard |
Roles were clearly defined to ensure accountability, with planned role rotation for cross-learning.
- Week 1: Setup & planning
- Week 2: AI model integration
- Week 3: Security & tamper detection
- Week 4: Dashboard & alerts
- Week 5: Testing, attack simulation, demo prep
Focus was on working software, not slides.
✔ Real-world problem with social impact ✔ Strong cybersecurity alignment ✔ Edge-based (not cloud-dependent) ✔ Live demo-ready ✔ Clear execution within 25 days
SentinAI shows how AI + Cybersecurity can:
- Prevent incidents instead of just recording them
- Protect surveillance systems from attacks
- Improve emergency response during the critical Golden Hour
This project was built under the CSI Core Projects Initiative 2026, emphasizing:
- Real execution
- Weekly accountability
- Industry-relevant outcomes
SentinAI is not just a model or a dashboard — it is a working, secure surveillance system designed for real-world deployment.