維持・管理／検査技術・診断 (8)

[VI-779] Automatic Extraction of Deformations by Deep Learning and AI

*渡辺 貞之¹、篠原 秀明¹、米丸 陸海¹、安部 正道³、久保 英里奈³、今永 侑希³、小西 真治²、伊藤 聡² (1. メトロレールファシリティーズ、2. 東京地下鉄、3. 計測検査)

Keywords:subway, tunnel, Automatic extraction, crack, Water leakage

The tunnel of Tokyo Metro is developing the development by image photography. Heretofore, it has been possible to trace the deformation from the image by manually judging the type of deformation such as cracking or water leakage. Therefore, there is a need to reduce costs due to the difference in quality due to individual differences and the time required for preparation. For the purpose of quality improvement and cost reduction, we examined the automatic extraction of distortion from images by using Deep Learning and AI technology.