README.md

SimNP: Learning Self-Similarity Priors Between Neural Points

Official PyTorch implementation of the paper "SimNP: Learning Self-Similarity Priors Between Neural Points", ICCV 2023.

Christopher Wewer¹ Eddy Ilg² Bernt Schiele¹ Jan Eric Lenssen¹

¹Max Planck Institute for Informatics, ²Saarland University

Project Page | Paper

Requirements

Install required packages

Make sure you have up-to-date NVIDIA drivers supporting CUDA 11.3

Run

conda env create -f environment.yml
conda activate simnp

Usage

1. Clone the repository git clone git@github.com:Chrixtar/SimNP.git.

2. Download the [dataset].

3. Download [pretrained model weights] and extract them into the repository.

4. Install requirements conda env create -f environment.yml.

5. (Optional) Setup options in options.

6. (Optional) Run training script with python train.py --yaml=[options_path].

7. Run inference script with

   python predict.py --experiment [path_to_experiment_output_directory] --date [datetime_of_experiment_run]
   

   , where an experiment is determined by a certain options file and the date is the datetime of the experiment run.

Citation

@inproceedings {wewer2023simnp,
    booktitle = {ICCV},
    title = {SimNP: Learning Self-Similarity Priors Between Neural Points},
    author = {Wewer, Christopher and Ilg, Eddy and Schiele, Bernt and Lenssen, Jan Eric},
    year = {2023},
}