Abstract: Unmanned Aerial Vehicles (UAVs), commonly known as drones, have become increasingly prevalent across various industries, including agriculture, surveillance, and cinematography. However, this proliferation also raises concerns related to privacy, security, and safety, necessitating effective drone detection systems. This project focuses on developing a robust drone object detection system using the YOLOv8 (You Only Look Once) model, a state-of-the-art deep learning algorithm renowned for its real-time performance and high accuracy. The dataset for this project, comprising diverse drone images, was collected from the Roboflow platform and meticulously annotated using the PASCAL VOC XML format. The YOLOv8 model was trained on this annotated dataset, optimizing key parameters to minimize loss functions. Evaluation metrics such as precision, recall, and mean Average Precision (mAP) were employed to assess the model's performance, demonstrating a high accuracy rate in detecting drones under various conditions. Additionally, a user-friendly web application was developed using the Flask framework, allowing users to upload images or videos for real-time drone detection. This comprehensive approach, encompassing data collection, model training, evaluation, and application development, showcases the system's potential in enhancing security and safety measures in scenarios requiring drone monitoring and control.
Keywords: Machine learning, deep learning, NLP, GenAi, syntaxlibrary, C, java, javascripts.
|
DOI:
10.17148/IJARCCE.2024.13910
