The system employs an advanced AI object detection framework enhanced with proprietary Virtual Observation and Noise Filtering technologies that improve multi-object tracking robustness and accelerate processing speed on edge devices. Our innovative Dimension Transformer performs complex spatial calculations integrating camera parameters with geographic coordinates to enable precise crowd density monitoring. With documented precision rates of 93.6% and recall rates of 91.0%, the AI model efficiently processes data directly on edge devices ranging from smartphones to dedicated AIoT systems and drones, enabling accurate, real-time crowd density alerts with minimal infrastructure requirements.
A specific designed payload for drone makes the above technologies enabled to the drone-based wide area coverage. Key functionalities include people counting across specific lines, total number of people on screen, and area-based density estimation. The solution incorporates real-time alerts when crowd density exceeds 5 people per square meter - the established safety threshold. The system's mobility and easy setup enable rapid deployment at various locations, from footbridges to museums, markets, and promenades, providing frontline
officers with critical data for operational decisions during large-scale events.