Abstract
Unmanned Aerial Vehicles (UAVs), commonly known as drones, have emerged as versatile tools with applications spanning surveillance, aerial photography, agriculture, and disaster response. However, their increasing presence raises security concerns, necessitating robust detection systems. This research explores the development of a real-time drone detection system using the state-of-the-art YOLOv5 algorithm. This paper presents a detailed methodology, comparative analysis, and results demonstrating the efficacy of our approach in enhancing security measures against UAV threats. Current detection technologies encompass a range of approaches, including computer vision, machine learning, radar systems, and acoustic sensors. Traditional methods often rely on rule-based algorithms or handcrafted features, exhibiting limited scalability and adaptability to dynamic environments. In this research paper, we present a novel approach to drone detection utilizing the YOLOv5, a powerful object detection algorithm, with OpenCV, a versatile computer vision library. The heart of our system lies in a meticulously curated dataset containing 1440 images, showcasing a diverse array of drones. Each image tells a unique story, helping our system learn to recognize drones of different types and sizes. Our methodology involves a detailed process of training YOLOv5 using the dataset, carefully splitting the data into training, validation, and testing sets, and setting up a real-time detection system using OpenCV. The system not only identifies drones but also issues warnings when a drone is detected within or near a specified area.
Keywords: Computer Vision, Custom Dataset, Drone Detection, Real-Time Detection, YOLOv5 Algorithm.