https://github.com/pvnieo/snk
Pytorch code for "SNK: Shape Non-rigid Kinematics" - NeurIPS 2023
https://github.com/pvnieo/snk
pytorch
Last synced: about 2 months ago
JSON representation
Pytorch code for "SNK: Shape Non-rigid Kinematics" - NeurIPS 2023
- Host: GitHub
- URL: https://github.com/pvnieo/snk
- Owner: pvnieo
- License: gpl-3.0
- Created: 2023-09-27T06:43:00.000Z (over 1 year ago)
- Default Branch: master
- Last Pushed: 2024-05-17T13:54:26.000Z (about 1 year ago)
- Last Synced: 2025-03-26T21:47:38.236Z (2 months ago)
- Topics: pytorch
- Language: Python
- Homepage: https://arxiv.org/abs/2403.06804
- Size: 27.3 KB
- Stars: 6
- Watchers: 4
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# :racehorse: SNK :racehorse:
[](https://arxiv.org/abs/2403.06804)
Pytorch code for "SNK: Shape Non-rigid Kinematics" - NeurIPS 2023
> Zero-shot non-rigid correspondence
---
## :construction_worker: Installation
This implementation requires Python >= 3.7. Install dependencies using pip:
```bash
pip install -r requirements.txt
```
## :book: Usage
The full SNK pipeline will be available soon. Stay tuned!
Meanwhile, we release the code for the prism decoder (Section 4.2 in the paper). The prism decoder is a neural network that takes as input a source shape, and a latent code, and deform the source shape to a target shape that corresponds to the latent code. The code is available in the `prism_decoder` folder.
## :chart_with_upwards_trend: Results
If you wish to report our results, we have summarized them below. Our method is referred to as **SNK**. `X on Y` indicates that the method was trained on dataset `X` and tested on dataset `Y`.
- **Near Isometric Shape Matching**: We provide results on the FAUST (F), Scape (S), and SHREC (SH) datasets. We used the remeshed versions. We report the mean geodesic error, following the protocol used in all deep functional map papers. Our method is **Zero-Shot**, i.e. does not require any training on the train set, and is applied directly on the test set.
| Method | F | S | SH |
| --- | --- | --- | --- |
| **SNK** | 1.8 | 4.7 | 5.8 |
- **Non-Isometric Shape Matching**: We provide results on the SMAL dataset. We report the mean geodesic error, following the same protocol as in all the deep functional maps papers. Our method is **Zero-Shot**, i.e. does not require any training on the train set, and is applied directly on the test set.
| Method | SMAL |
| --- | --- |
| **SNK** | 9.1 |
## :mortar_board: Citation
If you find this work useful in your research, please consider citing:
```bibtex
@inproceedings{attaiki2023snk,
title={Shape Non-rigid Kinematics (SNK): A Zero-Shot Method for Non-Rigid Shape Matching via Unsupervised Functional Map Regularized Reconstruction},
author={Souhaib Attaiki and Maks Ovsjanikov},
booktitle={Advances in Neural Information Processing Systems},
year={2023}
}
```