DETECTION, TRACKING AND CLASSIFICATION OF ROGUE DRONES USING COMPUTER VISION

Abstract

The increase in the volume of UAVs has been rapid in the past few years. The utilization of drones has increased considerably in the military and commercial setups, with UAVs of all sizes, shapes, and types being used for various applications, from recreational flying to purpose-driven missions. This development has come with challenges and has been identified as a potential source of operational disruptions leading to various security complications, including threats to Critical Infrastructures (CI). Thus, the need for developing fully autonomous antiUAV Defense Systems (AUDS) hasn't been more imminent than today. To attenuate and nullify the threat posed by the UAVs, either deliberately or otherwise, this paper presents the holistic design and operational prototype of drone detection technology based on visual detection using Digital Image Processing (DIP) and Machine Learning (ML) to detect, track and classify drones accurately. The proposed system uses a background-subtracted frame difference technique for detecting moving objects partnered with a Pan-Tilt tracking system powered by Raspberry Pi to track the moving object. The identification of moving objects is made by a Convolutional Neural Network (CNN) system called the YOLO v4-tiny ML algorithm. The novelty of the proposed system lies in its accuracy, effectiveness with low-cost sensing equipment, and better performance compared to other alternatives. Along with ease of operations, combining the system with other systems like RADAR could be a real game-changer in detection technology. The experimental validation of the proposed technology was justified in various tests in an uncontrolled outdoor environment (in the presence of clouds, birds, trees, rain, etc.), proving to be equally effective in all the situations yielding high-quality results.