open access

https://hdl.handle.net/2115/100421

Crack detection using CNN with video images of railway tunnel central passage

英語

Image processing

2026-07-01

Proceedings of the 10th International Symposium on Reliability Engineering and Risk Management (ISRERM 2026)

Improving the efficiency of crack inspections is a challenge in the maintenance and management of concrete railway structures. Conventional close-up visual inspections have been problematic due to errors and oversights arising from subjective judgment, as well as the effort required for recording them. Therefore, in this study, we attempted to detect cracks by video recording on the side walls of the central passage of a railway tunnel. Given the on-site conditions, we built a video recording system using a tablet and lighting equipment and considered shooting parameters such as resolution, frame rate, and illuminance. In the future, we will convert videos into still images and identify cracks, aiming to improve the efficiency of inspections and advance the system through automatic diagnosis

会議発表論文

Version of Record