BEVDiffLoc

[IROS 25] BEVDiffLoc: End-to-End LiDAR Global Localization in BEV View based on Diffusion Model

Visualization

Environment

• python 3.10

• pytorch 2.1.2

• cuda 12.1

source install.sh

Dataset

We support the Oxford Radar RobotCar and NCLT datasets right now.

The data of the Oxford and NCLT dataset should be organized as follows:

data_root
├── 2019-01-11-14-02-26-radar-oxford-10k
│   ├── velodyne_left
│   │   ├── xxx.bin
│   │   ├── xxx.bin
│   ├── gps
│   │   ├── gps.csv
│   │   ├── ins.csv
│   ├── velodyne_left.timestamps
│   ├── merge_bev (Data prepare)
│   │   ├── xxx.png
│   │   ├── xxx.png
│   ├── merge_bev.txt (Data prepare)
├── Oxford_pose_stats.txt
├── train_split.txt
├── valid_split.txt

Data prepare

• Oxford&NCLT: We use merge_nclt.py and merge_oxford.py to generate local scenes for data augmentation.

Run

Download the pretrained ViT model

We initialize BEVDiffLoc's feature learner with DINOv2.

Train

accelerate launch --num_processes 1 --mixed_precision fp16 train_bev.py

Test

python test_bev.py

Citation

If you find this work helpful, please consider citing:

@article{wang2025bevdiffloc,
  title={BEVDiffLoc: End-to-End LiDAR Global Localization in BEV View based on Diffusion Model},
  author={Wang, Ziyue and Shi, Chenghao and Wang, Neng and Yu, Qinghua and Chen, Xieyuanli and Lu, Huimin},
  journal={arXiv preprint arXiv:2503.11372},
  year={2025}
}

Acknowledgement

We appreciate the code of DiffLoc and BEVPlace++ they shared.