ログイン
Language:

WEKO3

  • トップ
  • ランキング
To
lat lon distance
To

Field does not validate



インデックスリンク

インデックスツリー

メールアドレスを入力してください。

WEKO

One fine body…

WEKO

One fine body…

アイテム

  1. 学位論文
  2. 学位論文

小物体検出のためのYOLOv8の改良およびその保育環境への応用

https://doi.org/10.18997/0002001056
https://doi.org/10.18997/0002001056
28bf47c5-d555-4118-a367-66fa43c9447f
名前 / ファイル ライセンス アクション
sei_k_497.pdf sei_k_497.pdf (5.9 MB)
アイテムタイプ 学位論文 = Thesis or Dissertation(1)
公開日 2024-11-21
資源タイプ
資源タイプ識別子 http://purl.org/coar/resource_type/c_db06
資源タイプ doctoral thesis
タイトル
タイトル Small Object Detection Based on Improved YOLOv8 and Its Application to Nursery Environment
言語 en
タイトル
タイトル 小物体検出のためのYOLOv8の改良およびその保育環境への応用
言語 ja
言語
言語 jpn
著者 Supri Bin Amir,

× Supri Bin Amir,

en Supri Bin Amir,

Search repository
抄録
内容記述タイプ Abstract
内容記述 The aim of this study is to establish a general methodology for the detection of small objects, applying it to the nursery environment as an example. In the dynamic environment of a nursery, children’s safety is always a priority. One of the key challenges in nursery environments is that children always interact with a multitude of objects, such as toys, stationery, or small accessories, which can be critical for understanding the activities and interactions within the space that may pose a risk or potential danger. In this situation, developing monitoring technology based on object detection is important to increase safety in this environment. In order to simulate real conditions, we collected an image dataset from a nursery. We identify a variety of complex scenarios within the nursery environment, including objects of varying sizes, limited visual features, and partial object occlusion. This study focuses on the development of small object detection algorithm based on YOLOv8. Unfortunately, standard YOLOv8 doesn’t yield the best results when applied to a complex nursery environment. To enhance the detection accuracy of nursery objects using the YOLOv8 model, we first comprehensively investigate the impact of incorporating attention mechanisms into the YOLOv8 network architecture. The attention mechanism works by selectively paying attention to the salient parts of a visual feature while ignoring irrelevant ones. This addresses the limitations of the visual features of objects that are influenced by the quality of the image’s data or the object’s size in a nursery environment. Furthermore, our objective is to enhance the performance of the YOLOv8 in detecting small objects. The nursery environment has a large number of small objects with limited visual features, potentially leading to partial occlusion between objects during children’s activities within the space, such that the same object will vary in size. The standard YOLOv8 model encounters difficulties in accurately distinguishing between an actual object and a visually similar object or background. In an effort to address the issue of standard YOLOv8’s limited ability to detect small objects, we introduce the YOLOv8s-NE model. This model aims to improve the detection of small objects, specifically in nursery environments. Finally, we apply our proposed YOLOv8s-NE model in the nursery’s safety system. We specifically developed an algorithm to identify objects on top of the cabinet that could potentially fall and threaten children.
目次
内容記述タイプ TableOfContents
内容記述 1 Introduction| 2 Related Literature| 3 Analyzing the impact of integrating attention mechanisms into the YOLOv8 for the nursery dataset| 4 Enhancing Small Object Detection for Nursery Environment| 5 Discussion| 6 Conclusions
備考
内容記述タイプ Other
内容記述 九州工業大学博士学位論文 学位記番号:生工博甲第497号 学位授与年月日:令和6年9月25日
学位授与番号
学位授与番号 甲第497号
学位名
学位名 博士(工学)
学位授与年月日
学位授与年月日 2024-09-25
学位授与機関
学位授与機関識別子Scheme kakenhi
学位授与機関識別子 17104
学位授与機関名 九州工業大学
言語 ja
学位授与年度
内容記述タイプ Other
内容記述 令和6年度
出版タイプ
出版タイプ VoR
出版タイプResource http://purl.org/coar/version/c_970fb48d4fbd8a85
アクセス権
アクセス権 open access
アクセス権URI http://purl.org/coar/access_right/c_abf2
ID登録
ID登録 10.18997/0002001056
ID登録タイプ JaLC
戻る
0
views
See details
Views

Versions

Ver.1 2024-11-21 02:51:32.628038
Show All versions

Share

Share
tweet

Cite as

Other

print

エクスポート

OAI-PMH
  • OAI-PMH JPCOAR 2.0
  • OAI-PMH JPCOAR 1.0
  • OAI-PMH DublinCore
  • OAI-PMH DDI
Other Formats
  • JSON
  • BIBTEX
  • ZIP

コミュニティ

確認

確認

確認


Powered by WEKO3


Powered by WEKO3