NyxSentra

Inspiration

Urban areas have many CCTV cameras and street lights, but they mostly work in a reactive way, recording incidents after they happen. Many poorly lit zones still feel unsafe, especially for pedestrians and women. We wanted to explore how AI could make existing infrastructure intelligent and proactive in improving urban safety.

What it does

NyxSentraAI is an AI-powered urban safety system that analyzes CCTV video in real time to detect suspicious behavior such as abnormal loitering or panic movements. When a potential threat is detected, it generates alerts and increases nearby street-light brightness to deter suspicious activity and assist authorities.

How we built it

We used Python, OpenCV, and YOLOv8 for real-time person detection and behavior analysis, with MediaPipe for pose estimation. The backend was built using FastAPI/Flask with MongoDB, and a React-based dashboard was created to visualize alerts and safety insights.

Challenges we ran into

Handling real-time video processing efficiently was challenging. Another issue was reducing false positives in behavior detection. Low-light environments also made detection difficult and required additional optimization.

Accomplishments that we're proud of

We built a working prototype capable of detecting human activity, generating threat alerts, and demonstrating the concept of adaptive smart lighting and centralized monitoring.

What we learned

We gained experience in computer vision, real-time AI systems, and full-stack integration, and learned how AI can be applied to solve real-world urban safety problems.

What's next for NyxSentra

We plan to improve behavior prediction accuracy, develop crime heatmaps and predictive safety analytics, and integrate more deeply with IoT-based smart city infrastructure.