News:
- 🔥 December, 2024. CAD-Recode is state-of-the-art in three CAD reconstruction benchmarks:
DeepCAD
Fusion360
CC3D
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, Elona Dupont, Dimitrios Mallis, Kseniya Cherenkova, Anis Kacem, Djamila Aouada
Univesity of Luxembourg
https://arxiv.org/abs/2412.14042
CAD-Recode transforms point cloud to 3D CAD model in form of Python code (CadQuery library). CAD-Recode is trained upon Qwen2-1.5B, keeping original tokenizer, and adding a single additional linear layer. Model weights are hosted on HuggingFace. Our interactive demo in hosted on 🤗 ZeroGPU spaces (5 GPU minutes for free!).
In this repo we provide simple inference demo. Install python packages according to our Dockerfile and run demo.ipynb in jupyter.
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}
}