Score-Based Multibeam Point Cloud Denoising (AUV Symposium 2024)

[Paper] TODO

Installation

Recommended Environment

The code has been tested in the following environment:

Package	Version	Comment
Python	3.8
PyTorch	1.9.0
pytorch3d	0.5.0	Used to compute k nearest neighbors in the MBES point cloud.
Open3D	0.18.0	Used for baseline evaluations.

Install via Conda (PyTorch 1.9.0 + CUDA 11.1)

conda env create -f mbes_env.yml
conda activate mbes-score-denoise

Datasets

To create the dataset for training, please check out the mbes-cleaning repository.

Training and Evaluation

Both training and testing are performed using train_orig_mbes.py file. For testing, add the --test flag.

Example training command:

python train_orig_mbes.py \
       --raw_data_root <path to folder with raw data patches> \
       --gt_root <path to folder with with draping results ground truth patches>

Example testing command:

python train_orig_mbes.py \
       --raw_data_root <path to folder with raw data patches> \
       --gt_root <path to folder with with draping results ground truth patches> \
       --test \
       --ckpt_path <path to the .pt checkpoint for evaluation>

Please find tunable parameters in the script.

Citation

#TODO