Drone Tracking and Object Detection By YOLO And CNN

Abstract

This thesis focuses on the utilization of YOLO (You Only Look Once) and CNN (Convolutional Neural Network) for real-time drone detection. The study explores the fundamentals of YOLO and CNN, including their working principles and mathematical equations for object detection. A specialized dataset comprising diverse drone images is collected and meticulously annotated for training the models. Evaluation of the trained models is conducted using established metrics such as mAP and IoU. The results highlight the models' performance compared to baseline approaches, demonstrating their strengths and limitations. A comprehensive workflow for drone detection employing YOLO and CNN is presented, encompassing dataset collection, model training, evaluation, and deployment stages. This research contributes valuable insights to the field of drone detection and offers prospects for future enhancements and applications.