The Proceedings of the Conference on Information, Intelligence and Precision Equipment : IIP
Online ISSN : 2424-3140
2022
Session ID : IIP1R3-I12
Conference information

3D coordinates estimation and 3D object placement system from monocular image
Yoshihiro HAGIGARA*Shunki MORIUCHI
Author information
CONFERENCE PROCEEDINGS RESTRICTED ACCESS

Details
Abstract

In recent years, AR games for smartphones have become popular, and the overall market for AR games is expanding. As a result, the demand for the development of systems for AR games is expected to increase in the future. In this research, we aim to develop a system that can be used in AR games for smartphones. Using a monocular image depth estimation method based on cycle GAN, which exist as prior research, rather than images from special cameras such as stereo cameras, we identify the coordinates of actual structures in the game space and develop a 3D object placement system based on these coordinates using Unity. As a result, For the depth estimation method, the accuracy of the RMSE was better than that of previous studies. In addition, we were able to construct a 3D object placement system using depth images and segmentation on Unity Engine.

Content from these authors
© 2022 The Japan Society of Mechanical Engineers
Previous article Next article
feedback
Top