|
ヤマザキ ヨシアキ
Yoshiaki Yamazaki
山崎 芳昭 所属 理工学部 総合理工学科 データサイエンス学環 職種 教授 |
|
| 言語種別 | 英語 |
| 発行・発表の年月 | 2025/01 |
| 形態種別 | 学術雑誌 |
| 査読 | 査読あり |
| 標題 | Validation of the Application of Object Detection Technology Using YOLOv9 for Rescue Robots
in Disaster Environments |
| 執筆形態 | 共著 |
| 掲載誌名 | Proceedings of the Joint Symposium of The Thirtieth International Symposium onArtificial Life and Robotics(AROB 30th 2025) |
| 掲載区分 | 国外 |
| 出版社・発行元 | International Society of Artificial Life and Robotics |
| 巻・号・頁 | ISBN978-4-9913442-1-3,pp.569-574 |
| 総ページ数 | 6 |
| 担当区分 | 最終著者,責任著者 |
| 著者・共著者 | Anyu Ishizaka, Jehun Seo, Yoshiaki Yamazaki |
| 概要 | Many disaster sites are difficult environments for people to enter due to the high risk of secondary disasters, such as
fire or hazardous chemicals. For this reason, object detection technology using machine learning has attracted attention as a means of assisting information gathering in rescue operations and understanding the disaster situation. However, there are issues where detection accuracy is reduced due to environmental factors such as multiple detection targets, lighting conditions, perspective distortion of images, camera angle, blurring of images and changes in contrast. In this study, the detection performance of hazardous material labels in disaster environments was verified using You Only Look Once. Furthermore, the influence of the dataset structure, such as the presence or absence of background images, on the detection accuracy was clarified. This demonstrated the effectiveness of the model, with the aim of improving its practicality in rescue operations |
| researchmap用URL | https://researchmap.jp/yamazakiyoshiaki/ |