FOSS4G-Asia 2023 Seoul

Development of deep learning model for road mapping
11-30, 16:30–17:00 (Asia/Seoul), Gallery 3

Recently, the rapid advancement of autonomous driving-related technologies has led to swift progress in the development of various sensing devices and application methods. In particular, plans to produce highly precise maps to support this technology are being promoted, and research and data production projects related to this are being conducted continuously.

On the other hand, the nature of maps, which must maintain a certain level of accuracy, makes them difficult to update frequently. This is due to the substantial manpower and financial resources required not only for initial production but also for continuous updating of the maps.

I believe that the most fitting solution to this problem is AI technology. It has already been actively applied in other fields and can provide both speed and cost-effectiveness. However, since the deep learning models currently available have limitations in terms of accuracy when producing maps, a separate AI model that can guarantee this accuracy must be developed. In this study, we aim to introduce a method that can automatically update road maps by enhancing previously announced deep learning models and integrating various spatial data.

Mar. 2023 - Present: Assistant Professor, Drone and GIS Engineering, Namseoul University
Jan. 2023 – Feb. 2023: Postdoctoral researcher, KICT(Korea Institute of Construction Technology)
Mar. 2017 – Aug. 2022: Ph.D., School of Civil & Environmental Engineering, Yonsei University
Sep. 2009 – Aug. 2011: Master Degree, School of Civil & Environmental Engineering, Yonsei University
Mar. 2007 – Aug. 2009: Bachelor of Civil & Environmental Engineering, Yonsei University