NetraRakshak- An AI Approach for Indian CCTV Networks
Link to Live Project: https://github.com/Dakshified/NetraRakshak--An-AI-Approach-for-Indian-CCTV-Networks
Public safety in India demands smarter surveillance. While CCTV networks are widely deployed in schools, malls, metros, buses, and temples, most systems remain reactive — useful only after an incident occurs. Existing AI-based surveillance detects weapons only when they are clearly visible. In reality, attackers often hide knives, pistols, or sharp tools in bags and pockets, revealing them only at the last moment. By then, prevention is nearly impossible. Another limitation is false alarms, where harmless objects such as umbrellas, bottles, or phones are flagged as weapons. Moreover, Western-trained models focus mainly on guns and knives, overlooking India-specific threats like rods, sickles, cutters, and glass bottles.
Our solution, NetraRakshak, is an India-first AI surveillance system that identifies not just weapons, but also the actions leading to violence. The pipeline begins with CCTV or IP camera feeds, processed by YOLOv8 to detect people, bags, and objects. DeepSORT tracks individuals across frames, while MediaPipe monitors hand movements. If a hand enters a bag or pocket, the extraction logic activates, watching for new objects. These are classified using MobileNet, distinguishing weapons from harmless items. At the same time, behavior analysis checks for threatening motions such as arm raises or thrusts. All signals are combined into a risk score, and if thresholds are exceeded, the system sends real-time alerts with annotated video clips to security staff.
NetraRakshak is affordable and scalable. It runs on laptops and low-cost edge devices such as Jetson Nano or Raspberry Pi, making it suitable for schools and institutions of all sizes. With temporal logic, synthetic data augmentation for India-specific weapons, and human-in-loop verification, the system ensures accuracy while reducing false positives.
NetraRakshak is a step toward proactive, ethical, and India-focused AI surveillance — an intelligent eye that protects communities by seeing what others miss.
This build was uploaded as a hackathon project
