Real Time Drone Detection Using YOLO and Coordinate Attention Integrated GELAN-E Models

Abstract

In the recent scenario drones have play important role in war, aerial surveillance, illegal spying, maintain public safety in sensitive areas, logistics and environmental observation. To prevent the unauthorized drone, use the drone detection in real time. In real time one of the difficult tasks in object detection is drone detection due to visibility conditions, complex backgrounds, weather condition, tiny visible things and similar objects that are difficult to identify. Our study proposes a real-time drone identification object using YOLOv8, YOLOv11, and GELAN-E integrated with Coordinate Attention deep learning models, that enhanced processes and interpretability of object detection techniques. By integrating the GELAN-E integrated with Coordinate Attention YOLO is known as YOLOv8n-CA-GE YOLOv11n-CA-GE, the study significantly improves   performance, particularly for small and hidden drones. The proposed drone detection modelYOLOv8n-CA-GE with an accuracy of precision 0.962, a recall of 0.917, mAP@0.5score of 0.956 andmAP@0.5:0.95 score of 0.755, YOLOv11n-CA-GEwith an accuracy of precision 0.962, a recall of 0.930, mAP@0.5 score of 0.966andmAP@0.5:0.95 score of 0.773. For all weather and day conditions. In addition to visual interpretability, enhances drone detection performance in real-time.