DeFillet

KO | EN

gitlite — search

#typescript #golang #rust #macos #java #windows #awesome #cli #kubernetes #vue #pytorch #security #mcp #ios

(★ 92)

The code of paper "DeFillet: Detection and Removal of Fillet Regions in Polygonal CAD Models" , ACM Transactions on Graphics (SIGGRAPH 2025)

DeFillet 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<h1 align="center">DeFillet: Detection and Removal of Fillet Regions in Polygonal CAD Models</h1>

[//]: # ( <a href="https://github.com/xiaowuga/DeFillet">)

[//]: # (<img src="https://img.shields.io/badge/Poster-PDF-purple?logo=adobeacrobatreader&logoColor=white" alt="Poster">)

[//]: # ( </a>)

<a target="_blank" href="https://xiaowuga.github.io/">Jing-En Jiang</a>
 ·
 <a target="_blank" href="https://github.com/hanxiaowang00">Hanxiao Wang</a>
 ·
 <a target="_blank" href="https://zikai1.github.io/">Mingyang Zhao</a>
 ·
 <a target="_blank" href="https://sites.google.com/site/yandongming/">Dong-Ming Yan</a>
 ·
 <a target="_blank" href="https://xk.qust.edu.cn/info/1041/4695.htm">Shuangmin Chen</a>
 .
 <a target="_blank" href="https://irc.cs.sdu.edu.cn/~shiqing/index.html">Shiqing Xin</a>
 .
 <a target="_blank" href="https://faculty.sdu.edu.cn/tuzhanghe/en/index.htm">Changhe Tu</a>
 .
 <a target="_blank" href="https://engineering.tamu.edu/cse/profiles/Wang-Wenping.html">Wenping Wang</a>

![](https://raw.githubusercontent.com/xiaowuga/DeFillet/main/asset/teaser.gif)

This repository contains the official implementation of our SIGGRAPH 2025 paper "DeFillet: Detection and Removal of Fillet Regions in Polygonal CAD Models".

**Please give a star and cite if you find this repo useful.**

### 📢 Windows Version Released

The [**Windows Executables**](https://github.com/xiaowuga/DeFillet/releases/tag/v1.0) are now live. 💻
Feel free to test and provide feedback. 📝

> ⚠️ **Note:** For evaluation only.

## 💻 Platform
- Windows 11
- CLion2024.1.2 +  Visual Stdio 2022
- Intel(R) Core i9-13900K

## 📦 Dependence

The dependent libraries of our code includes:
- Eigen3 (3.4.0 or later): Solving linear equations.
- libigl (2.5.0 or later): Solving linear equations.
- Easy3D (2.5.2 or later): IO operation and data structure for mesh.
- CGAL (5.6 or later): Voronoi diagram computation.
- CLI11 (2.4.0 or later): Command line.
- nlohmann_json (3.11.3 or later): Read json file.

For Eigen3, libigl, CGAL, CLI11 and nlohmann_json, we recommend using [vcpkg](https://github.com/microsoft/vcpkg) to install them.
```shell
# Eigen3
vcpkg install Eigen3:x64-windwos
# libigl
vcpkg install libigl:x64-windwos
# CGAL
vcpkg install cgal:x64-windwos
# cli11
vcpkg install cli11:x64-windows
# nlohmann_json
vcpkg install nlohmann_json:x64-windows
```

For Easy3D, please visit the [Prof.Nan's repository](https://github.com/LiangliangNan/Easy3D) to obtain it and follow the instructions for installation. You need to set the `${Easy3D_DIR}` environment variable to the directory that contains the `Easy3DConfig.cmake` file.

## 🛠️ How to Build

Building our code in CLion:
```
# File -> Setting -> Build, Execution, Deployment -> CMake -> CMake Option :
-DCMAKE_TOOLCHAIN_FILE=${YOUR_VCPKG_INSTALL_PATH}/scripts/buildsystems/vcpkg.cmake
```
Making sure that your following settings are correct:
- Toolchains : `Visual Stdio`
- Architecture : `amd64`
- Build Type : `release`

## 🚀 Usage

### 1.detector_cli

```shell
detector_cli.exe -c detector_config.json
```

The executable target `detector_cli.exe` takes a `.ply` mesh file as input and outputs the fillet segmentation results and intermediate results to a specified output folder.
You can modify the `detector_config.json` file to change the input and output folder paths as well as algorithm parameters (configured based on the details in the paper).

### 2.removal_cli

```shell
removal_cli.exe -c removal_config.json
```
The executable target `detector_cli.exe` accepts two types of input: a `.ply` mesh file and a JSON-format segmentation result file (generated by `detector_cli.exe` itself). It then outputs both the defilleted shape and intermediate data to a user-specified output directory.
You can modify the `removal_config.json` file to change the input and output folder paths as well as algorithm parameters (configured based on the details in the paper).

## 💡 Tips
- The code for **distinguishing cylinders from fillets** mentioned in the paper has not been released due to stability issues. Given the high geometric similarity between these features, separating them is essentially a semantic segmentation problem, suggesting that deep learning might be a more effective solution.
- Setting `num_patches = -1` in `detector_config.json` enables the use of the global Voronoi diagram. Due to its superior stability, this approach yields better detection performance for models that differ from the examples shown in Fig. 8 of the paper.
- The primary parameters to tune in `detector_config.json` are `epsilon` and `radius_thr`. A larger `epsilon` value induces over-segmentation, which improves robustness when dealing with low-resolution meshes. This results in more stable normals during the removal process, ultimately yielding superior results.

## 📂 Dataset
You can download some test results for DeFillet dataset from my [Google Drive](https://drive.google.com/file/d/1_jlt2WN2XOxvy3manOtUdhNnhcgL4IUH/view?usp=drive_link). Note that in some of these examples, the models were first split into multiple components based on dihedral angles to detect fillets separately, and then stitched back together. 
This approach was adopted to achieve better algorithmic performance.

## 📝 Citation
If you make use of our work, please cite our paper:

```bibtex
@article{jiang2025defillet,
    author = {Jiang, Jing-En and Wang, Hanxiao and Zhao, Mingyang and Yan, Dong-Ming and Chen, Shuangmin and Xin, Shiqing and Tu, Changhe and Wang, Wenping},
    title = {DeFillet: Detection and Removal of Fillet Regions in Polygonal CAD Models},
    journal={ACM Transactions on Graphics (TOG)},
    publisher = {Association for Computing Machinery},
    year = {2025},
    address = {New York, NY, USA},
    volume = {44},
    number = {4},
    issn = {0730-0301},
    url = {https://doi.org/10.1145/3731166},
    doi = {10.1145/3731166} 
}
```

## 📧 Maintaince

If any problem, please contact me via <xiaowuga@gmail.com>.

## 📜 License
DeFillet is under AGPL-3.0, so any downstream solution and products (including cloud services) that include DeFillet code inside it should be open-sourced to comply with the AGPL conditions. For learning purposes only and not for commercial use. If you want to use it for commercial purposes, please contact us first.

DeFillet

DeFillet 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

DeFillet

DeFillet 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README