HeightFormer: Explicit Height Modeling without Extra Data for Camera-only 3D Object Detection in Bird's Eye View

arXiv:2307.13510v2 Announce Type: replace
Abstract: Vision-based Bird's Eye View (BEV) representation is an emerging perception formulation for autonomous driving. The core challenge is to construct BEV space with multi-camera features, which is a one-to-many ill-posed problem. Diving into all previous BEV representation generation methods, we found that most of them fall into two types: modeling depths in image views or modeling heights in the BEV space, mostly in an implicit way. In this work, we propose to explicitly model heights …

3d object 3d object detection abstract arxiv autonomous autonomous driving bird challenge construct core cs.cv data detection driving extra features modeling object perception representation space type view vision