MinkLoc3D: Point Cloud Based Large-Scale Place Recognition

For improved version of MinkLoc3D method see: MinkLoc3Dv2

Paper: MinkLoc3D: Point Cloud Based Large-Scale Place Recognition 2021 IEEE Winter Conference on Applications of Computer Vision (WACV) arXiv

Supplementary material

Jacek Komorowski

Warsaw University of Technology

What's new

• [2021-09-29] Updated version of MinkLoc3D code is released. Changes include: optimization of training and evaluation pickles generation process; code updated to work with recent version of Pytorch and MinkowskiEngine.
• [2022-03-03] Improved vresion of MinkLoc3D method is released: MinkLoc3Dv2

Our other projects

• MinkLoc++: Lidar and Monocular Image Fusion for Place Recognition (IJCNN 2021): MinkLoc++
• Large-Scale Topological Radar Localization Using Learned Descriptors (ICONIP 2021): RadarLoc
• EgonNN: Egocentric Neural Network for Point Cloud Based 6DoF Relocalization at the City Scale (IEEE Robotics and Automation Letters April 2022): EgoNN
• Improving Point Cloud Based Place Recognition with Ranking-based Loss and Large Batch Training (2022): MinkLoc3Dv2

Introduction

The paper presents a learning-based method for computing a discriminative 3D point cloud descriptor for place recognition purposes. Existing methods, such as PointNetVLAD, are based on unordered point cloud representation. They use PointNet as the first processing step to extract local features, which are later aggregated into a global descriptor. The PointNet architecture is not well suited to capture local geometric structures. Thus, state-of-the-art methods enhance vanilla PointNet architecture by adding different mechanism to capture local contextual information, such as graph convolutional networks or using hand-crafted features. We present an alternative approach, dubbed MinkLoc3D, to compute a discriminative 3D point cloud descriptor, based on a sparse voxelized point cloud representation and sparse 3D convolutions. The proposed method has a simple and efficient architecture. Evaluation on standard benchmarks proves that MinkLoc3D outperforms current state-of-the-art.

Citation

If you find this work useful, please consider citing:

@INPROCEEDINGS{9423215,
  author={Komorowski, Jacek},
  booktitle={2021 IEEE Winter Conference on Applications of Computer Vision (WACV)}, 
  title={MinkLoc3D: Point Cloud Based Large-Scale Place Recognition}, 
  year={2021},
  volume={},
  number={},
  pages={1789-1798},
  doi={10.1109/WACV48630.2021.00183}}

Environment and Dependencies

Code was tested using Python 3.8 with PyTorch 1.9.1 and MinkowskiEngine 0.5.4 on Ubuntu 20.04 with CUDA 10.2. Note: CUDA 11.1 is not recommended as there are some issues with MinkowskiEngine 0.5.4 on CUDA 11.1.

The following Python packages are required:

• PyTorch (version 1.9.1)
• MinkowskiEngine (version 0.5.4)
• pytorch_metric_learning (version 1.0 or above)
• tensorboard
• pandas

Modify the PYTHONPATH environment variable to include absolute path to the project root folder:

export PYTHONPATH=$PYTHONPATH:/home/.../MinkLoc3D

Datasets

MinkLoc3D is trained on a subset of Oxford RobotCar and In-house (U.S., R.A., B.D.) datasets introduced in PointNetVLAD: Deep Point Cloud Based Retrieval for Large-Scale Place Recognition paper (link). There are two training datasets:

• Baseline Dataset - consists of a training subset of Oxford RobotCar
• Refined Dataset - consists of training subset of Oxford RobotCar and training subset of In-house

For dataset description see PointNetVLAD paper or github repository (link).

You can download training and evaluation datasets from here (alternative link).

Before the network training or evaluation, run the below code to generate pickles with positive and negative point clouds for each anchor point cloud. NOTE: Training and evaluation pickles format has changed in this release of MinkLoc3D code. If you have created these files using the previous version of the code, they must be removed and re-created.

cd generating_queries/ 

# Generate training tuples for the Baseline Dataset
python generate_training_tuples_baseline.py --dataset_root <dataset_root_path>

# Generate training tuples for the Refined Dataset
python generate_training_tuples_refine.py --dataset_root <dataset_root_path>

# Generate evaluation tuples
python generate_test_sets.py --dataset_root <dataset_root_path>

<dataset_root_path> is a path to dataset root folder, e.g. /data/pointnetvlad/benchmark_datasets/. Before running the code, ensure you have read/write rights to <dataset_root_path>, as training and evaluation pickles are saved there.

Training

To train MinkLoc3D network, download and decompress the dataset and generate training pickles as described above. Edit the configuration file (config_baseline.txt or config_refined.txt). Set dataset_folder parameter to the dataset root folder. Modify batch_size_limit parameter depending on available GPU memory. Default limit (=256) requires at least 11GB of GPU RAM.

To train the network, run:

cd training

# To train minkloc3d model on the Baseline Dataset
python train.py --config ../config/config_baseline.txt --model_config ../models/minkloc3d.txt

# To train minkloc3d model on the Refined Dataset
python train.py --config ../config/config_refined.txt --model_config ../models/minkloc3d.txt

Pre-trained Models

Pretrained models are available in weights directory

• minkloc3d_baseline.pth trained on the Baseline Dataset
• minkloc3d_refined.pth trained on the Refined Dataset

Evaluation

To evaluate pretrained models run the following commands:

cd eval

# To evaluate the model trained on the Baseline Dataset
python evaluate.py --config ../config/config_baseline.txt --model_config ../models/minkloc3d.txt --weights ../weights/minkloc3d_baseline.pth

# To evaluate the model trained on the Refined Dataset
python evaluate.py --config ../config/config_refined.txt --model_config ../models/minkloc3d.txt --weights ../weights/minkloc3d_refined.pth

Results

MinkLoc3D performance (measured by Average Recall@1%) compared to state-of-the-art:

Trained on Baseline Dataset

Method	Oxford	U.S.	R.A.	B.D
PointNetVLAD [1]	80.3	72.6	60.3	65.3
PCAN [2]	83.8	79.1	71.2	66.8
DAGC [3]	87.5	83.5	75.7	71.2
LPD-Net [4]	94.9	96.0	90.5	89.1
EPC-Net [5]	94.7	96.5	88.6	84.9
SOE-Net [6]	96.4	93.2	91.5	88.5
NDT-Transformer [7]	97.7
MinkLoc3D (our)	97.9	95.0	91.2	88.5

Trained on Refined Dataset

Method	Oxford	U.S.	R.A.	B.D
PointNetVLAD [1]	80.1	94.5	93.1	86.5
PCAN [2]	86.4	94.1	92.3	87.0
DAGC [3]	87.8	94.3	93.4	88.5
LPD-Net [4]	94.9	98.9	96.4	94.4
SOE-Net [6]	96.4	97.7	95.9	92.6
MinkLoc3D (our)	98.5	99.7	99.3	96.7

1. M. A. Uy and G. H. Lee, "PointNetVLAD: Deep Point Cloud Based Retrieval for Large-Scale Place Recognition", 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
2. W. Zhang and C. Xiao, "PCAN: 3D Attention Map Learning Using Contextual Information for Point Cloud Based Retrieval", 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
3. Q. Sun et al., "DAGC: Employing Dual Attention and Graph Convolution for Point Cloud based Place Recognition", Proceedings of the 2020 International Conference on Multimedia Retrieval
4. Z. Liu et al., "LPD-Net: 3D Point Cloud Learning for Large-Scale Place Recognition and Environment Analysis", 2019 IEEE/CVF International Conference on Computer Vision (ICCV)
5. L. Hui et al., "Efficient 3D Point Cloud Feature Learning for Large-Scale Place Recognition", preprint arXiv:2101.02374 (2021)
6. Y. Xia et al., "SOE-Net: A Self-Attention and Orientation Encoding Network for Point Cloud based Place Recognition", 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
7. Z. Zhou et al., "NDT-Transformer: Large-scale 3D Point Cloud Localisation Using the Normal Distribution Transform Representation", 2021 IEEE International Conference on Robotics and Automation (ICRA)

• J. Komorowski, "MinkLoc3D: Point Cloud Based Large-Scale Place Recognition", Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), (2021)

License

Our code is released under the MIT License (see LICENSE file for details).