卷積神經網絡模型在無人機碰撞迴避數據集上的性能分析;Performance Evaluation of CNN Models for Drone Collision Avoidance Dataset

NCU Institutional Repository > 資訊電機學院 > 通訊工程學系碩士在職專班 > 博碩士論文 > Item 987654321/92916

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/92916

题名:	卷積神經網絡模型在無人機碰撞迴避數據集上的性能分析;Performance Evaluation of CNN Models for Drone Collision Avoidance Dataset
作者:	蘇慧敏;Su, Huei-Min
贡献者:	通訊工程學系在職專班
关键词:	深度學習;影像偵測;YOLO;Object detection
日期:	2023-12-26
上传时间:	2024-09-19 16:30:46 (UTC+8)
出版者:	國立中央大學
摘要:	人工智慧隨著互聯網上龐大的圖像數據不斷增加也以加速的速度演進。無人機（UAVs）也是一種近幾年來具有影響力的新趨勢。然而，在使用UAVs引發了一些安全問題，如與動態障礙物（鳥類、其他飛機或隨機拋擲的物體）的碰撞。這些問題錯綜複雜，有時無法通過最先進的算法來完全避免，對應用程序構成潛在威脅。為了應對這些安全問題，推出了一個開創性的項目，名為ColANet，這是一個新的碰撞視頻數據集，旨在為訓練最先進的機器學習算法提供基礎，以高效處理碰撞迴避問題。而ColANet的主要目標是使這些算法能夠有效應對碰撞迴避問題的複雜性。ColANet不僅有助於訓練各種神經網絡模型，還展示了測試其有效性的便利性。這項研究具體針對使用YOLOv8技術進行UAV碰撞避免性能分析。實驗結果顯示，YOLOv8大幅提高了效率，為UAV碰撞避免系統的改進提供了有價值的見解。隨著UAV在各個行業中發揮著重要作用，這項研究的結果成為一個有價值的參考，強調了YOLOv8在顯著提高UAV碰撞避免系統效率和安全性方面的潛力。 ;Artificial intelligence is evolving rapidly alongside the exponential growth of image data on the Internet. Unmanned Aerial Vehicles (UAVs) have also emerged as a influential trend in recent years. However, the use of UAVs has raised safety concerns, particularly regarding collisions with dynamic obstacles such as birds, other aircraft, or randomly thrown objects. These issues are complex and sometimes cannot be entirely avoided even with state-of-the-art algorithms, posing potential threats to applications. To address these safety concerns, a groundbreaking initiative called ColANet has been introduced. ColANet is a novel collision video dataset designed to serve as a foundation for training state-of-the-art machine learning algorithms to efficiently handle collision avoidance problems. The primary objective of ColANet is to empower these algorithms to effectively address the complexity of collision avoidance problems. ColANet not only facilitates the training of various neural network models but also showcases the ease of testing their effectiveness. This research specifically focuses on the performance analysis of UAV collision avoidance using the YOLOv8 technology. The experimental results demonstrate that YOLOv8 significantly improves efficiency, providing valuable insights for enhancing UAV collision avoidance systems. As UAVs play a crucial role in various industries, the findings of this study serve as a valuable reference, emphasizing the potential of YOLOv8 in significantly improving the efficiency and safety of UAV collision avoidance systems.
显示于类别:	[通訊工程學系碩士在職專班 ] 博碩士論文

文件中的档案:

档案	描述	大小	格式	浏览次数
index.html		0Kb	HTML	38	检视/开启

在NCUIR中所有的数据项都受到原著作权保护.

社群 sharing

数据加载中.....