Reproduce “Voxel Transformer for 3D Object Detection” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jiageng Mao, Yujing Xue, Minzhe Niu, Haoyue Bai, Jiashi Feng, Xiaodan Liang, Hang Xu, Chunjing Xu
Voxel Transformer (VoTr) replaces sparse 3D convolutional backbones with selective local and dilated attention, allowing each query voxel to use features from a limited set of nearby and more distant voxels. On Waymo Open, VoTr-TSD reaches 74.95 vehicle 3D mean average precision on LEVEL 1 validation, versus 74.06 for reimplemented PV-RCNN. On KITTI, VoTr-SSD runs slower than SECOND.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.