The popularity of unmanned aerial vehicles (UAVs), also referred to as drones, is growing quickly, raising questions about their misuse as they become more widely available. This escalating trend underscores the urgency for effective drone detection mechanisms to prevent unlawful intrusions and ensure public safety. To bridge the existing gap in technology, There’s a deliberate attempt to develop sophisticated deep-learning models capable of accurately recognizing drones of varying shapes and forms, including those that resemble birds. In our quest to address this challenge, we have adopted a comprehensive approach by utilizing the YOLOv3 object detection model alongside OpenCV. Through meticulous training on a diverse dataset encompassing images of drones, including bird-like variations, we have achieved an impressive accuracy rate of 96.98% at 0.5 mAP. Moreover, our integration of DeepSORT with YOLO enhances the system’s tracking capabilities, enabling efficient monitoring and tracking of identified drones. In conclusion, our endeavor not only fills a crucial void in current drone detection technology but also underscores our commitment to safeguarding against potential threats posed by drone misuse. Through continuous advancements in deep-learning methodologies and innovative integration techniques, we aim to establish robust frameworks for automated UAV detection, ensuring the security and integrity of airspace operations.
UAV Object Detection and Tracking in Video using YOLOv3 and Deepsort
2024-04-22
5993782 byte
Conference paper
Electronic Resource
English
Traffic Object Detection and Distance Estimation Using YOLOv3
SAE Technical Papers | 2022
|