Abstract: Recent progress on multi-modal 3D object detection has featured BEV (Bird-Eye-View) based fusion, which effectively unifies both LiDAR point clouds and camera images in a shared BEV space.
Abstract: This paper proposes a new 3D multi-object tracker to more robustly track objects that are temporarily missed by detectors. Our tracker can better leverage object features for 3D Multi-Object ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results