Applying image recognition technology by convolutional neural networks to mountain plot images

Nomura, Masahiro  ; Tamura, Fumihiko   ; Shimada, Taihei ; Yamamoto, Masanobu   ; Furusawa, Masashi*; Sugiyama, Yasuyuki*; Hara, Keigo*; Hasegawa, Katsushi*; Omori, Chihiro*; Yoshii, Masahito*

Image recognition using a convolutional neural network (CNN) has been used in a wide range of fields and has produced excellent results. If this image recognition technology is used effectively, it should be possible to obtain information from an image equal to or more than the information that a person can obtain from an image. At J-PARC, researchers with specialized knowledge obtain beam information needed to adjust the equipment from an image called mountain plot. In this study, we applied the image recognition technology by using CNN to this mountain plot image, and tried to obtain the information about the beam necessary for adjustment. As a result, we were able to obtain more information than is currently available by using the image recognition technology. In the future, we plan to adjust the equipment based on the information actually obtained from the image recognition technology and confirm its effectiveness