CoCa3D

This repository contains the official PyTorch implementation of

Collaboration Helps Camera Overtake LiDAR in 3D Detection
Yue Hu, Yifan Lu, Runsheng Xu, Weidi Xie, Siheng Chen, Yanfeng wang
Presented at CVPR 2023

LiDAR detection v.s. collaborative camera perception

Main idea

Abstract: Camera-only 3D detection provides an economical solution with a simple configuration for localizing objects in 3D space compared to LiDAR-based detection systems. However, a major challenge lies in precise depth estimation due to the lack of direct 3D measurements in the input. Many previous methods attempt to improve depth estimation through network designs, e.g., deformable layers and larger receptive fields. This work proposes an orthogonal direction, improving the camera-only 3D detection by introducing multi-agent collaborations. Our preliminary results show a potential that with sufficient collaboration, the camera might overtake LiDAR in some practical scenarios.

Features

• Dataset Support

  • DAIR-V2X
  • OPV2V+
  • V2X-SIM

• SOTA collaborative perception method support

  • Where2comm [Neurips2022]
  • V2VNet [ECCV2020]
  • DiscoNet [NeurIPS2021]
  • V2X-ViT [ECCV2022]
  • When2com [CVPR2020]
  • Late Fusion
  • Early Fusion

• Visualization

  • BEV visualization
  • 3D visualization

Quick Start

Install

Please refer to the INSTALL.md for detailed documentations.

Download dataset DAIR-V2X

1. Download raw data of DAIR-V2X.
2. Download complemented annotation from dair-v2x-c-complemented.

Train your model

We adopt the same setting as OpenCOOD which uses yaml file to configure all the parameters for training. To train your own model from scratch or a continued checkpoint, run the following commonds:

python opencood/tools/train.py --hypes_yaml ${CONFIG_FILE} [--model_dir  ${CHECKPOINT_FOLDER}]

Arguments Explanation:

• hypes_yaml: the path of the training configuration file, e.g. opencood/hypes_yaml/second_early_fusion.yaml, meaning you want to train an early fusion model which utilizes SECOND as the backbone. See Tutorial 1: Config System to learn more about the rules of the yaml files.
• model_dir (optional) : the path of the checkpoints. This is used to fine-tune the trained models. When the model_dir is given, the trainer will discard the hypes_yaml and load the config.yaml in the checkpoint folder.

Test the model

Before you run the following command, first make sure the validation_dir in config.yaml under your checkpoint folder refers to the testing dataset path, e.g. opv2v_data_dumping/test.

python opencood/tools/inference.py --model_dir ${CHECKPOINT_FOLDER} --fusion_method ${FUSION_STRATEGY} --save_vis_n ${amount}

Arguments Explanation:

• model_dir: the path to your saved model.
• fusion_method: indicate the fusion strategy, currently support 'early', 'late', 'intermediate', 'no'(indicate no fusion, single agent), 'intermediate_with_comm'(adopt intermediate fusion and output the communication cost).
• save_vis_n: the amount of saving visualization result, default 10

The evaluation results will be dumped in the model directory.

Acknowledgements

Thank for the excellent cooperative perception codebases OpenCOOD and CoPerception.

Thank for the excellent cooperative perception datasets DAIR-V2X, OPV2V and V2X-SIM.

Thanks for the insightful previous works in cooperative perception field. Where2comm(NeruIPS22), CoAlign(ICRA23), V2VNet(ECCV20), When2com(CVPR20), Who2com(ICRA20), DiscoNet(NeurIPS21), V2X-ViT(ECCV2022), STAR(CoRL2022), CoBEVT(CoRL2022).

Contact

If you have any problem with this code, please feel free to contact [email protected].

Citation

If you find this code useful in your research then please cite

@inproceedings{CoCa3D:23,
  author    = {Yue Hu, Yifan Lu, Runsheng Xu, Weidi Xie, Siheng Chen, Yanfeng Wang},
  title     = {Collaboration Helps Camera Overtake LiDAR in 3D Detection},
  booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year      = {2023}
}