Imaging-sonar-based Object Recognition Utilizing Object's Yaw Angle Estimation with Deep Learning Technique
- Imaging-sonar-based Object Recognition Utilizing Object's Yaw Angle Estimation with Deep Learning Technique
- YU, SON CHEOL; Minsung Sung; MEUNGSUK, LEE; BYEONGJIN, KIM
- Date Issued
- IFAC(International Federation of Automatic Control)
- This paper proposes a method to recognize underwater target objects and estimate their yaw angle using an imaging sonar. First, a fast sonar simulator generated template images of the target objects for various viewpoints. Next, a generative adversarial network (GAN) generated a semantic map by segmenting the real sonar image for reliable recognition. Then, matching the template image and the semantic map identies the target object and yaw angle of the object. We veried the proposed method through indoor water tank experiments by installing objects and acquiring images from various angles. The proposed method can provide relative position and bearing information of a sensing platform so that it can be applied to various algorithms such as pose control and navigation.
- Article Type
- 2020 IFAC World Congress, page. 1 - 6, 2020-07-11
- Files in This Item:
- There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.