Center Focusing Network for Real-Time LiDAR Panoptic Segmentation

Official code for CFNet

Center Focusing Network for Real-Time LiDAR Panoptic Segmentation, Xiaoyan Li, Gang Zhang, Boyue Wang, Yongli Hu, Baocai Yin. (https://openaccess.thecvf.com/content/CVPR2023/papers/Li_Center_Focusing_Network_for_Real-Time_LiDAR_Panoptic_Segmentation_CVPR_2023_paper.pdf) Accepted by CVPR2023

NEWS

• [2023-02-24] CFNet is accepted by CVPR 2023
• [2022-11-17] CFNet achieves the 63.4 PQ and 68.3 mIoU on the SemanticKITTI LiDAR Panoptic Segmentation Benchmark with the inference latency of 43.5 ms on a single NVIDIA RTX 3090 GPU.

1 Dependency

CUDA>=10.1
Pytorch>=1.5.1
PyYAML@5.4.1
scipy@1.3.1

2 Training Process

2.1 Installation

cd pytorch_lib
python setup.py install

2.2 Prepare Dataset

Please download the SemanticKITTI dataset to the folder ./data and the structure of the folder should look like:

./data
    ├── SemanticKITTI
        ├── ...
        └── dataset/
            ├──sequences
                ├── 00/         
                │   ├── velodyne/
                |   |	├── 000000.bin
                |   |	├── 000001.bin
                |   |	└── ...
                │   └── labels/ 
                |       ├── 000000.label
                |       ├── 000001.label
                |       └── ...
                ├── 08/ # for validation
                ├── 11/ # 11-21 for testing
                └── 21/
                    └── ...

And download the object bank on the SemanticKITTI to the folder ./data and the structure of the folder should look like:

./data
    ├── object_bank
        ├── bicycle
        ├── bicyclist
        ├── car
        ├── motorcycle
        ├── motorcyclist
        ├── other-vehicle
        ├── person
        ├── truck

2.3 Training Script

python3 -m torch.distributed.launch --nproc_per_node=8 train.py --config config/config_mvfcev2ctx_sgd_wce_fp32_lossv2_single_newcpaug.py

3 Evaluate Process

python3 -m torch.distributed.launch --nproc_per_node=8 evaluate.py --config config/config_mvfcev2ctx_sgd_wce_fp32_lossv2_single_newcpaug.py --start_epoch 0 --end_epoch 47

Citations

@inproceedings{licfnet2023,
  author={Li, Xiaoyan and Zhang, Gang and Wang, Boyue and Hu, Yongli and Yin, Baocai},
  booktitle={2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, 
  title={Center Focusing Network for Real-Time LiDAR Panoptic Segmentation}, 
  year={2023},
  volume={},
  number={},
  pages={13425-13434},
  doi={10.1109/CVPR52729.2023.01290}
}