Adaptive Feature Fusion for Cooperative Perception using LiDAR Point Clouds [WACV2023][paper]

Installation

Please refer to the OpenCOOD repository for setup and installation.

Some installation commands used in this research

Install bbx NMS calculation CUDA version:

python3 opencood/utils/setup.py build_ext --inplace

Setup:

python3 setup.py develop --user

Install spconv:

e.g.

pip3 install spconv-cu113

Training

python3 opencood/tools/train.py --hypes_yaml opencood/hypes_yaml/point_pillar_spatialcooper.yaml

Evaluation

Before you run the following command, first make sure the validation_dir in config.yaml under your checkpoint folder.

• Testing dataset path: opv2v_data_dumping/test.
• Culver City dataset path: opv2v_data_dumping/test_culver_city.

python3 opencood/tools/inference.py --model_dir opencood/logs/point_pillar_spatialcooper/ --fusion_method intermediate

Qualitative Results on OPV2V Dataset

Single Vehicle Perception v.s. Cooperative Perception

Qualitative Results on CODD Dataset

Cooperative perception for vehicle and pedestrian detection

Acknowledgement

• Xu, R., Xiang, H., Xia, X., Han, X., Li, J. and Ma, J., 2022, May. Opv2v: An open benchmark dataset and fusion pipeline for perception with vehicle-to-vehicle communication. In 2022 International Conference on Robotics and Automation (ICRA) (pp. 2583-2589). IEEE.