SGLoc: Scene Geometry Encoding for Outdoor LiDAR Localization
-
python 3.8.16
-
pytorch 1.11.0
-
cuda 11.3
source install.sh
We support the Oxford Radar RobotCar and NCLT datasets right now.
We also use PQEE to enhance the Oxford and provide the corrected pose as QEOxford.
The data of the Oxford, QEOxford and NCLT dataset should be organized as follows:
data_root
├── 2019-01-11-14-02-26-radar-oxford-10k
│ ├── xxx.bin
│ ├── xxx.bin
├── Oxford_pose_stats.txt
├── train_split.txt
├── valid_split.txt
We use SPVNAS for data preprocessing (just used for training) and generate corresponding planar masks. You need to download the code for SPVNAS and run the data_prepare.py we provided within it.
python train.py --dataset_folder xxx
python test.py --dataset_folder xxx --resume_model checkpoint_epoch_xxx.pth
The models of SGLoc on Oxford, QEOxford, and NCLT can be downloaded here.
We appreciate the code of PosePN and STCLoc they shared.
@inproceedings{li2023sgloc,
title={SGLoc: Scene Geometry Encoding for Outdoor LiDAR Localization},
author={Li, Wen and Yu, Shangshu and Wang, Cheng and Hu, Guosheng and Shen, Siqi and Wen, Chenglu},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={9286--9295},
year={2023}
}