改良型 YOLOv8 模型於 nVIDIA 高效能運算器之即時無人機影像偵測與追蹤;Modified YOLOv8 Model for Real-Time Drone Image Detection and Tracking with nVIDIA HPC

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Communication Engineering > Electronic Thesis & Dissertation > Item 987654321/95369

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/95369

Title:	改良型 YOLOv8 模型於 nVIDIA 高效能運算器之即時無人機影像偵測與追蹤;Modified YOLOv8 Model for Real-Time Drone Image Detection and Tracking with nVIDIA HPC
Authors:	游知欣;Yu, Chih-Hsin
Contributors:	通訊工程學系
Keywords:	無人機偵測;視覺偵測;物件偵測;注意力機制;YOLOv8;Drone Detection;Visual Detection;Object Detection;Attention Mechanism;YOLOv8
Date:	2024-08-16
Issue Date:	2024-10-09 16:43:15 (UTC+8)
Publisher:	國立中央大學
Abstract:	偵測不同機種的無人機對於確保安全合法使用和防範潛在風險至關重要，本文利用視覺偵測的方法對無人機進行偵測，以 YOLOv8 物件偵測模型對六種不同型態的無人機進行影像辨識、分類。根據無人機在影像上為小物件的特性透過注意力機制對模型進行改良，使模型更加注重部分的影像訊息，抓取小物件的特徵；另外一方面，因應無人機的移動特性，在模型中加入可變形卷積，透過可變形卷積中加入偏移量的機制，使模型影像更能準確的偵測無人機。實驗結果表明，本文提出的在原始 YOLOv8 架構中加入注意力機制以及可變形卷積地模型比原始 YOLOv8 模型有更高的精確率。;Detecting different drone models is crucial for ensuring safe andlegal use while mitigating potential risks.In this paper employs visual de tection using the YOLOv8 model to identify and classify six drone types.To enhance small object detection, attention mechanisms are integrated to focus on specific image details. Additionally, deformable convolutionsare included to address drone movement, improving detection accuracy.Results indicate that the proposed model, with attention mechanisms anddeformable convolutions, surpasses the original YOLOv8 model in pre cision.
Appears in Collections:	[Graduate Institute of Communication Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	65	View/Open

社群 sharing

Loading...