An AI-powered surveillance system built to automate video monitoring, detect anomalies in real time, and generate actionable insights via a smart dashboard.
Manual review of CCTV footage is time-consuming, error-prone, and often leads to delayed detection of threats or incidents. Our solution automates surveillance using computer vision and deep learning to enable intelligent, real-time monitoring.
- 🎯 Real-time activity detection and anomaly classification
- 📦 Centralized storage of activity logs, alerts, and summaries
- 📄 Auto-generated PDF reports using Groq LLM
- 📊 Interactive Streamlit dashboard for visualization
- 🧠 Graph-based relationship tracking (Neo4j) for event mapping
- Python, PyTorch, FFmpeg, OpenCV
- YOLOv8 + DeepSORT for object tracking
- 3D ResNet-18 for spatiotemporal action recognition
- Whisper (planned) for audio event detection
- Streamlit + Neo4j for dashboard and storage
| Feature | Domains | Use Case Examples |
|---|---|---|
| 🧠 Edge Deployment | Smart Cities, Industry | Jetson Nano detects traffic violations or PPE violations on-site |
| 🔊 Audio Analysis (Whisper) | Hospitals, Transport | Detect glass breaking, shouting, or cries for help |
| 🛎️ Auto Alert Routing | Airports, Hostels | Route alerts to closest guard based on location |
| 🎥🔊 Sensor Fusion | Banks, Events, Homes | Combine audio-visual cues for intelligent decisions |
- Trisha Sharma
- Anahita Bhandari
- Ananya Manan
Built with 💡 by Team Data Mavericks, Manipal University Jaipur