SimMOD

Implementation of SimMOD: A Simple Baseline for Multi-Camera 3D Object Detection. (AAAI 2023)

Arxiv

Installation

Check installation for installation.

Data Preparation

Check data_preparation for preparing the nuScenes dataset.

Getting Started

To train SimMOD with 8 GPUs, run:

bash tools/dist_train.sh $CONFIG 8

For evaluation, use:

bash tools/dist_test.sh $CONFIG $CKPT 8 --eval=bbox

We provide the pretrained models for SimMOD.

Method	Pretrain	mAP	NDS	Log	Weights
SimMOD-r50	ImageNet	33.1	42.7	log	model
SimMOD-r101	ImageNet	34.9	43.1	log	model
SimMOD-r101	FCOS3D	37.0	45.4	log	model
SimMOD-r101	NuImg	37.6	46.1	log	model

Acknowledgement

This project is mainly based on DETR3D. Thanks for their great work.

If you find this repo useful for your research, please consider citing the paper:

@article{zhang2022simple,
  title={A Simple Baseline for Multi-Camera 3D Object Detection},
  author={Zhang, Yunpeng and Zheng, Wenzhao and Zhu, Zheng and Huang, Guan and Zhou, Jie and Lu, Jiwen},
  journal={arXiv preprint arXiv:2208.10035},
  year={2022}
}