https://github.com/filapro/cad-recode
CAD-Recode: Reverse Engineering CAD Code from Point Clouds
https://github.com/filapro/cad-recode
cad cadquery llm pytorch qwen2
Last synced: about 1 year ago
JSON representation
CAD-Recode: Reverse Engineering CAD Code from Point Clouds
- Host: GitHub
- URL: https://github.com/filapro/cad-recode
- Owner: filaPro
- License: other
- Created: 2024-12-18T17:27:07.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-03-19T16:13:24.000Z (about 1 year ago)
- Last Synced: 2025-04-09T09:05:23.247Z (about 1 year ago)
- Topics: cad, cadquery, llm, pytorch, qwen2
- Language: Jupyter Notebook
- Homepage:
- Size: 179 KB
- Stars: 93
- Watchers: 4
- Forks: 10
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.md
- License: LICENSE.md
Awesome Lists containing this project
README
## CAD-Recode: Reverse Engineering CAD Code from Point Clouds
:hugs: Model [v1](https://huggingface.co/filapro/cad-recode), [v1.5](https://huggingface.co/filapro/cad-recode-v1.5) :hugs: ZeroGPU Space [v1.5](https://huggingface.co/spaces/filapro/cad-recode) :hugs: Dataset [v1](https://huggingface.co/datasets/filapro/cad-recode), [v1.5](https://huggingface.co/datasets/filapro/cad-recode-v1.5)
**News**:
* :rocket: March, 2025. We update CAD-Recode to v1.5. More details in [changelog](CHANGELOG.md).
* :fire: December, 2024. CAD-Recode is state-of-the-art in three CAD reconstruction benchmarks:
DeepCAD [](https://paperswithcode.com/sota/cad-reconstruction-on-deepcad?p=cad-recode-reverse-engineering-cad-code-from)
Fusion360 [](https://paperswithcode.com/sota/cad-reconstruction-on-fusion-360-gallery?p=cad-recode-reverse-engineering-cad-code-from)
CC3D [](https://paperswithcode.com/sota/cad-reconstruction-on-cc3d?p=cad-recode-reverse-engineering-cad-code-from)
This repository contains an implementation of CAD-Recode, a 3D CAD reverse engineering method introduced in our paper:
> **CAD-Recode: Reverse Engineering CAD Code from Point Clouds**
> [Danila Rukhovich](https://github.com/filaPro),
> [Elona Dupont](https://scholar.google.com/citations?user=i9J6YFMAAAAJ),
> [Dimitrios Mallis](https://scholar.google.com/citations?user=Gfc5ZXoAAAAJ),
> [Kseniya Cherenkova](https://scholar.google.com/citations?user=VepvFBkAAAAJ),
> [Anis Kacem](https://scholar.google.com/citations?user=K3EWusMAAAAJ),
> [Djamila Aouada](https://scholar.google.com/citations?user=WBmJVSkAAAAJ)
> Univesity of Luxembourg
> https://arxiv.org/abs/2412.14042
### Inference Demo
CAD-Recode transforms point cloud to 3D CAD model in form of Python code ([CadQuery](https://github.com/CadQuery/cadquery) library). CAD-Recode is trained upon Qwen2-1.5B, keeping original tokenizer, and adding a single additional linear layer.
In this repo we provide simple inference demo. Install python packages according to our [Dockerfile](Dockerfile) and run [demo.ipynb](demo.ipynb) in jupyter.
### Citation
If you find this work useful for your research, please cite our paper:
```
@article{rukhovich2024cadrecode,
title={CAD-Recode: Reverse Engineering CAD Code from Point Clouds},
author={Danila Rukhovich, Elona Dupont, Dimitrios Mallis, Kseniya Cherenkova, Anis Kacem, Djamila Aouada},
journal={arXiv preprint arXiv:2412.14042},
year={2024}
}
```