Citation

Kuroki M, Tanaka S, Muto K, Inou H. Trans. Soc. Automot. Eng. Jpn. 2022; 53(4): 802-807.

Vernacular Title

自動運転での点群を用いたAI 物体検出における判定根拠の可視化

Copyright

(Copyright © 2022, Society of Automotive Engineers of Japan)

DOI

10.11351/jsaeronbun.53.802

PMID

unavailable

Abstract

While AI has been applied to object detection in the fields of ADAS and AD, a lot of research of Explainable AI, which is used to explain evidences of object detection results, has been conducted. Explainable AI has been mainly developed for 2D object detection and it has been rarely applied for 3D object detection. Therefore, we adapted the existing method of Explainable AI for 2d object detection to 3D one using point cloud and we report its result.

===

自動運転の物体検出にAIが活用されつつあり、同技術の検出結果を説明するための「説明可能AI」の研究が世の中で取り組まれている。同技術は主に2D物体技術を中心に開発が進んでいるものの、3D物体技術への応用例はまだ少ない。今回、2D向け説明可能AI技術を3D物体検出に応用したのでその結果を報告する。

Language: ja

Keywords

Explainable AI; Image processing/Information processing; Intelligent vehicle; Safety