README.md

BEVHeight++: Toward Robust Visual Centric 3D Object Detection

Lei Yang · Tao Tang · Jun Li · Kun Yuan · Peng Chen · Li Wang · Yi Huang · Lei Li · Xinyu Zhang · Kaicheng Yu

BEVHeight++ is a new vision-based 3D object detector specially designed for both roadside and vihicle-side scenarios. On popular 3D detection benchmarks of roadside cameras, BEVHeight++ surpasses all previous vision-centric methods by a significant margin. In terms of the ego-vehicle scenario, our BEVHeight++ also possesses superior over depth-only methods.

Getting Started

0. Installation as BEVDet

1. Prepare nuScenes Dataset

Download nuScenes dataset from official website.

ln -s [nuScenes-dataset-root] ./data/nuscenes

2. Prepare infos for nuScenes dataset.

python tools/create_data_bevheight_plus.py

3. Train BEVHeight++ with 8 GPUs

# stage 1: 
bash tools/dist_train.sh configs/bevheight_plus/bevheight_plus-r50-depth-cbgs-first-stage.py 8

# stage 2:
mv [PTH_PATH_OF_STAGE_1] pretrained_model/epoch_20_ema.pth
bash tools/dist_train.sh configs/bevheight_plus/bevheight_plus-r50-depth-cbgs.py 8

4. Eval BEVHeight++ with 8 GPUs

bash tools/dist_test.sh configs/bevheight_plus/bevheight_plus-r50-depth-cbgs.py [PTH_PATH_OF_STAGE_2] 8 --eval mAP

Results on NuScenes Test Set.

Model	Backbone	mAP	mATE	mASE	mAOE	mAVE	mAAE	NDS	Config	Download
BEVHeight++	V2-99	0.529	0.441	0.258	0.358	0.295	0.142	0.614	config	model / log

Acknowledgment

This project is not possible without the following codebases.

• BEVDet
• mmdetection3d

Citation

If you use BEVHeight++ in your research, please cite our work by using the following BibTeX entry:

@article{yang2023bevheight++,
  title={Bevheight++: Toward robust visual centric 3d object detection},
  author={Yang, Lei and Tang, Tao and Li, Jun and Chen, Peng and Yuan, Kun and Wang, Li and Huang, Yi and Zhang, Xinyu and Yu, Kaicheng},
  journal={arXiv preprint arXiv:2309.16179},
  year={2023}
}