A Machine Vision Approach to Real-Time Drone Detection Using Gray Level Co-occurrence Matrix and Interactive GUI

Abstract

Unmanned aerial vehicles have witnessed a surge in popularity, prompting concerns regarding potential misuse, including smuggling, terrorist activities, and unauthorized entry into restricted airspace. Consequently, there is a pressing need to develop effective drone detection systems. This study presents a novel DDS framework that harnesses state-of-the-art neural techniques to achieve precise and efficient drone detection. The methodology leverages the Gray Level Co-occurrence Matrix algorithm for drone identification and classification based on their textural features, which has demonstrated effectiveness in distinguishing drones from other airborne objects, birds and various drone categories. The proposed framework underwent rigorous evaluation on a comprehensive dataset, exhibiting superior efficacy and precision in real-time scenarios compared to alternative DDS techniques. The robustness and effectiveness of this framework position it as a prime choice for securityoriented entities, particularly those operating in air force and military domains. Our experimental results reveal a commendable accuracy rate of 97%, affirming the reliability and precision of our framework in accurately identifying and detecting drones, surpassing recent models in the field. Additionally, our framework demonstrates a recall parameter of 98%, further underscoring its efficacy in real-world scenarios.