Search


Information
ICMST-Tohoku 2018
Oct. 23 - 26, 2018
Sendai, Japan
ICMST-Shenzhen 2016
Nov 1 - 4, 2016
Shenzhen, China
(EXIT THIS PAGE)
ICMST-Kobe 2014
Nov 2(Sun) - 5(Wed), 2014
Kobe, Japan
Nuclear Regulation Authority Outline of the New Safety Standards for Light Water Reactors for Electric Power Generation
For Public Comment
Outline of New Safety Standard (Design Basis)
For Public Comment
New Safety Standards (SA) Outline (Draft)
For Public Comment
Outline of New Safety Standard(Earthquake and Tsunami)(DRAFT)
Issues
 

Vol.10 No.2(Aug)
Vol.10 No.1(May)
Vol.9 No.4(Feb)
Vol.9 No.3(Nov)
< Other Issues

 

Occasional Topics
OTjapan Measures for Tsunami Striking Nuclear Power Station in Japan
Special Article: The Great Tohoku Earthquake (1)
OTjapan The Tragedy of “To Be” Principle in the Japanese Nuclear Industry
EJAMOT_CN3_Figure1_The_outside_view_of_CEFR OTChinaPlanning and Consideration on SFR R&D Activities in China
< All Occasional Topics

Featured Articles
EJAM7-3NT72 A New Mechanical Condition-based Maintenance Technology Using Instrumented Indentation Technique
EJAM7-3NT73 Survey robots for Fukushima Daiichi Nuclear Power Plant

JSM
Contacts
(EJAM): ejam@jsm.or.jp
(JSM): secretariat@jsm.or.jp
HP: http://www.jsm.or.jp
(in English)

 

Vol.17 No.2previousAA184
Academic Articles
Regular Paper Vol.17 No.2 (2025) p.15 - p.24
 

Method for Creating Large Datasets for Deep Learning to Improve Image Depth Accuracy

 

Masahiro MURAYAMA1,*,Yuki HARAZONO1,Hirotake ISHII1,Hiroshi SHIMODA1 and Yasuyoshi TARUTA2

 
1 Graduate School of Energy Science, Kyoto University, Sakyo-ku Yoshidahonmachi, Kyoto-shi, Kyoto 606-8501, Japan
2 Fugen Decommissioning Engineering Center, Japan Atomic Energy Agency, 3 Myojin-cho, Tsuruga-shi, Fukui, 914-8510, Japan
 
Abstract
High quality depth images are required for accurate 3D modeling of a facility. However, depth images captured using a typical commercially available RGB-D camera include much noise. Recently, methods using deep learning for depth enhancement have been developed. As described herein, we developed a novel method to create a high-quality dataset by generating high-quality depth images with pixel-wise depth enhancement, which is less affected by camera pose estimation errors. Furthermore, our method improves the quality of the entire dataset by post-processing suitable for our depth enhancement process. Comparison with the dataset created using the existing method showed that datasets created using the proposed method are suitable for training a network for depth enhancement. Depth images taken inside the Fugen Decommissioning Engineering Center are processed by a network trained on the dataset. The network completed the missing areas more correctly and removed the noise while maintaining the detail shapes.
 
Keywords
dataset creation, deep learning, depth image, noise removal
 
Full Paper: PDF
Article Information
Article history:
Received 18 November 2024
Accepted 9 June 2025