土木分野におけるIoT/AIのあり方（3)

[CS10-25] Evaluation of Characteristics on Tunnel Face Using Convolutional Neural Network

*鈴木 亮汰¹、松田 浩朗¹、兼松 亮¹ (1. 飛島建設株式会社)

Keywords: Evaluation of Tunnel Face, Image Processing, Deep Learning, Convolutional Neural Network

It is required to save labor and improve safety because the tunnel face evaluation and writing observation at the tunnel construction site are frequent and dangerous work. In order to solve these problems by automation, this research tried to evaluate the face characteristics from images. To process images, convolutional neural network (CNN) was used. In this paper, we proposed a classification CNN: predict evaluation points according to the format of observation, and a regression CNN: predict the continuous value of compressive strength.