Development of Web System for Detecting Cracks on Wall Surface using Deep Learning
| Project/Area Number |
20K12083
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 62020:Web informatics and service informatics-related
|
|---|
| Research Institution | Osaka Metropolitan University (2022)
Osaka City University (2020-2021) |
|---|
Principal Investigator |
YOSHIDA Daisuke 大阪公立大学, 大学院情報学研究科, 准教授 (00555344)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
川合 忠雄 大阪公立大学, 大学院工学研究科, 教授 (20177637)
瀧澤 重志 大阪公立大学, 大学院生活科学研究科, 教授 (40304133)
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Project Status |
Completed (Fiscal Year 2022)
|
| Budget Amount *help |
¥4,290,000 (Direct Cost: ¥3,300,000、Indirect Cost: ¥990,000)
Fiscal Year 2022: ¥260,000 (Direct Cost: ¥200,000、Indirect Cost: ¥60,000)
Fiscal Year 2021: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Fiscal Year 2020: ¥2,470,000 (Direct Cost: ¥1,900,000、Indirect Cost: ¥570,000)
|
|---|
| Keywords | 深層学習 / Webシステム / ドローン / ひび割れ検知 / 3次元モデル / ３次元モデル |
|---|
| Outline of Research at the Start |
本研究者はドローン等のICTを活用し，インフラ維持管理における応用研究を自治体と連携し進めてきた．これまでの研究成果として，深層学習を用いた護岸壁面のひび割れ検知プロトタイプを開発し，現在は検知結果の精度検証や，検知性能向上のための改良などを進めている．本研究では，このプロトタイプをWebシステムに実装することで広くシステムを利用可能にし，自治体との連携を通じて実用レベルでの有用性を明らかにする．
|
| Outline of Final Research Achievements |
The researchers have developed a detection program that utilizes deep learning for aerial images taken by small general-purpose drones used in the practice of local governments, and are researching and developing a web system that automatically detects cracks. We proceeded in cooperation with local governments that maintain and manage infrastructure (especially the Osaka Ports and Harbors Bureau). In the research, measurement experiments were carried out at several harbor quay walls and the performance of the detection system was evaluated. The current evaluation results show that the detection performance achieves an accuracy rate of 89% for cracks wider than 3mm (ranging from 3mm to several centimeters). However, in some field cases, it is possible to accurately detect cracks as small as 1mm, indicating significant room for improvement in the program. This reveals the need for further research and development.
|
| Academic Significance and Societal Importance of the Research Achievements |
本研究では，将来的に研究成果の他自治体・部局への横展開を想定しているため，高価で取り扱いが難しい産業用ドローンではなく，実際の業務で使用している比較的安価な小型汎用ドローンで得られる画像データを対象とし，深層学習を用いた物体検出手法の実装により，港湾施設点検に必要な検知性能を目指す点があげられる．また，この本研究を通じ，自治体職員のドローンの操縦を含めた計測データの活用がおこなえる人材育成についても同時に進めた．
|
Report
(4 results)
Research Products
(7 results)
