car-coder-907 / opencood

[ICRA 2022] An opensource framework for cooperative detection. Official implementation for OPV2V.

OpenCOOD

OpenCOOD is an Open COOperative Detection framework for autonomous driving. It is also the official implementation of the ICRA 2022 paper OPV2V.

News

03/17/2022: V2VNet is supported and the results/trained model are provided in the benchmark table.

03/10/2022: Results and pretrained weights for Attentive Fusion with compression are provided.

02/20/2022: F-Cooper now is supported and the results/traiend model can be found in the benchmark table.

01/31/2022: Our paper OPV2V: An Open Benchmark Dataset and Fusion Pipeline for Perception with Vehicle-to-Vehicle Communication has been accpted by ICRA2022!

09/21/2021: OPV2V dataset is public available: https://mobility-lab.seas.ucla.edu/opv2v/

Features

• Provide easy data API for the Vehicle-to-Vehicle (V2V) multi-modal perception dataset OPV2V

  It currently provides easy API to load LiDAR data from multiple agents simultaneously in a structured format and convert to PyTorch Tesnor directly for model use.

• Provide multiple SOTA 3D detection backbone

  It supports state-of-the-art LiDAR detector including PointPillar, Pixor, VoxelNet, and SECOND.

• Support most common fusion strategies

  It includes 3 most common fusion strategies: early fusion, late fusion, and intermediate fusion across different agents.

• Support several SOTA multi-agent visual fusion model

  It supports the most recent multi-agent perception algorithms (currently up to Sep. 2021) including Attentive Fusion, Cooper (early fusion), F-Cooper, V2VNet etc. We will keep updating the newest algorithms.

• Provide a convenient log replay toolbox for OPV2V dataset (coming soon)

  It also provides an easy tool to replay the original OPV2V dataset. More importantly, it allows users to enrich the original dataset by attaching new sensors or define additional tasks (e.g. tracking, prediction) without changing the events in the initial dataset (e.g. positions and number of all vehicles, traffic speed).

Data Downloading

All the data can be downloaded from google drive. If you have a good internet, you can directly download the complete large zip file such as train.zip. In case you suffer from downloading large fiels, we also split each data set into small chunks, which can be found in the directory ending with _chunks, such as train_chunks. After downloading, please run the following command to each set to merge those chunks together:

cat train.zip.parta* > train.zip
unzip train.zip

Installation

Please refer to data introduction and installation guide to prepare data and install OpenCOOD. To see more details of OPV2V data, please check our website.

Quick Start

Data sequence visualization

To quickly visualize the LiDAR stream in the OPV2V dataset, first modify the validate_dir in your opencood/hypes_yaml/visualization.yaml to the opv2v data path on your local machine, e.g. opv2v/validate, and the run the following commond:

cd ~/OpenCOOD
python opencood/visualization/vis_data_sequence.py [--color_mode ${COLOR_RENDERING_MODE}]

Arguments Explanation:

• color_mode : str type, indicating the lidar color rendering mode. You can choose from 'constant', 'intensity' or 'z-value'.

Train your model

OpenCOOD 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} [--show_vis] [--show_sequence]

Arguments Explanation:

• model_dir: the path to your saved model.
• fusion_method: indicate the fusion strategy, currently support 'early', 'late', and 'intermediate'.
• show_vis: whether to visualize the detection overlay with point cloud.
• show_sequence : the detection results will visualized in a video stream. It can NOT be set with show_vis at the same time.

The evaluation results will be dumped in the model directory.

Benchmark and model zoo

Results on OPV2V dataset ([email protected] for no-compression/ compression)

	Backbone	Fusion Strategy	Bandwidth (Megabit), before/after compression	Default Towns	Culver City	Download
Naive Late	PointPillar	Late	0.024/0.024	0.781/0.781	0.668/0.668	url
Cooper	PointPillar	Early	7.68/7.68	0.800/x	0.696/x	url
Attentive Fusion	PointPillar	Intermediate	126.8/1.98	0.815/0.810	0.735/0.731	url
F-Cooper	PointPillar	Intermediate	72.08/1.12	0.790/0.788	0.728/0.726	url
V2VNet	PointPillar	Intermediate	72.08/1.12	0.822/0.814	0.734/0.729	url
Naive Late	VoxelNet	Late	0.024/0.024	0.738/0.738	0.588/0.588	url
Cooper	VoxelNet	Early	7.68/7.68	0.758/x	0.677/x	url
Attentive Fusion	VoxelNet	Intermediate	576.71/1.12	0.864/0.852	0.775/0.746	url
Naive Late	SECOND	Late	0.024/0.024	0.775/0.775	0.682/0.682	url
Cooper	SECOND	Early	7.68/7.68	0.813/x	0.738/x	url
Attentive	SECOND	Intermediate	63.4/0.99	0.826/0.783	0.760/0.760	url
Naive Late	PIXOR	Late	0.024/0.024	0.578/0.578	0.360/0.360	url
Cooper	PIXOR	Early	7.68/7.68	0.678/x	0.558/x	url
Attentive	PIXOR	Intermediate	313.75/1.22	0.687/0.612	0.546/0.492	url

Note:

• We suggest using PointPillar as the backbone when you are creating your method and try to compare with our benchmark, as we implement most of the SOTA methods with this backbone only.
• We assume the transimssion rate is 27Mbp/s. Considering the frequency of LiDAR is 10Hz, the bandwidth requirement should be less than 2.7Mbp to avoid severe delay.
• A 'x' in the benchmark table represents the bandwidth requirement is too large, which can not be considered to employ in practice.

Tutorials

We have a series of tutorials to help you understand OpenCOOD more. Please check the series of our tutorials.

Citation

If you are using our OpenCOOD framework or OPV2V dataset for your research, please cite the following paper:

@inproceedings{xu2022opencood,
 author = {Runsheng Xu, Hao Xiang, Xin Xia, Xu Han, Jinlong Li, Jiaqi Ma},
 title = {OPV2V: An Open Benchmark Dataset and Fusion Pipeline for Perception with Vehicle-to-Vehicle Communication},
 booktitle = {2022 IEEE International Conference on Robotics and Automation (ICRA)},
 year = {2022}}

Also, under this LICENSE, OpenCOOD is for non-commercial research only. Researchers can modify the source code for their own research only. Contracted work that generates corporate revenues and other general commercial use are prohibited under this LICENSE. See the LICENSE file for details and possible opportunities for commercial use.

Future Plans

• Provide camera APIs for OPV2V
• Provide the log replay toolbox
• Implement F-Cooper
• Implement V2VNet
• Implement DiscoNet

Contributors

OpenCOOD is supported by the UCLA Mobility Lab. We also appreciate the great work from OpenPCDet, as part of our works use their framework.

Lab Principal Investigator:

• Dr. Jiaqi Ma (linkedin, UCLA Samueli)

Project Lead:

• Runsheng Xu (linkedin, github)
  
• Hao Xiang (linkedin, github)