simeco

KO | EN

gitlite — search

#python #hacktoberfest #go #rust #deep-learning #linux #nodejs #macos #windows #awesome #cli #awesome-list #security #open-source

(★ 22)

Learning generalizable shape completion with SIM(3) equivariance [NeurIPS 2025]

simeco 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# SIMECO: SIM(3)-Equivariant Shape Completion

**NeurIPS 2025**

[![Website](https://img.shields.io/badge/%F0%9F%A4%8D%20Project%20-Website-blue)](https://sime-completion.github.io)
[![arXiv](https://img.shields.io/badge/arXiv-PDF-b31b1b)](https://arxiv.org/abs/2509.26631)
[![Hugging Face Model](https://img.shields.io/badge/🤗%20Hugging%20Face%20-Model-yellow)](https://huggingface.co/wang2e2f/simeco)
[![Colab Demo](https://img.shields.io/badge/Colab-Demo-FF6F00?logo=googlecolab&logoColor=yellow)](https://colab.research.google.com/github/complete3d/simeco/blob/main/demo/demo.ipynb)
[![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](https://raw.githubusercontent.com/complete3d/simeco/main/LICENSE)

3D shape completion methods typically assume scans are pre-aligned to a canonical frame. When such alignment is absent in real data, performance collapses. SIMECO is a SIM(3)-equivariant network that delivers generalizable shape completion.

## Getting Started with the Demo 🤹‍♂️

You can run the demo in one of two ways:

- **Google Colab** (Recommended): No installation needed; get started instantly in the cloud.
- **Local Linux**: Set up the environment locally for GPU-accelerated inference.

### Google Colab ☁️

[<img src="https://colab.research.google.com/assets/colab-badge.svg" height="32"/>](https://colab.research.google.com/github/complete3d/simeco/blob/main/demo/demo.ipynb)

To run the demo on Google Colab, simply click the badge above. Follow the instructions and execute the cells sequentially with a GPU instance. Please note that setting up the environment and installing dependencies will take about 1-2 minutes.

### Local Linux 🖥️

> [!NOTE]
A CUDA-enabled GPU is required for local inference.

1. Clone the repository:

```bash
   git clone https://github.com/complete3d/simeco.git && cd simeco && git lfs pull
   ```

2. Create a conda environment with all dependencies (this will take approximately 5 minutes):

```bash
   . install.sh
   ```

3. Run the inference code and choose the desired transformation mode (e.g., `sim3`, `translation`, `scale` or `rotation`). The results will be saved in the `data/result` directory.

```bash
   python inference.py evaluate.single_file_path="demo/demo_data/pc/2a05d684eeb9c1cfae2ca9bb680dd18b.npy" evaluate.aug_mode="sim3" 
   ```
---

## 🚀 Usage:

### Training 🏋️‍♀️

* Start training using one of the two parallelization:

**Distributed Data Parallel (DDP):**
  
    ```bash
    # Replace device IDs with your own
    CUDA_VISIBLE_DEVICES=0,1 ./scripts/train_ddp.sh
    ```

**Data Parallel (DP):**
  
    ```bash
    # Replace device IDs with your own
    CUDA_VISIBLE_DEVICES=0,1 ./scripts/train_dp.sh
    ```

* Monitor training progress using TensorBoard:
  
  ```bash
  # Replace ${exp_name} with your experiment name (e.g., default)
  # Board typically available at http://localhost:6006
  tensorboard --logdir './output/${exp_name}/tensorboard'
  ```

### Evaluation 📊

* To evaluate a [pre-trained SIMECO model](./ckpt) using a single GPU:

```bash
  # Replace device IDs with your own
  CUDA_VISIBLE_DEVICES=0 ./scripts/test.sh
  ```

### Dataset 📂

We use the official PCN dataset. Please download the dataset from [PCN](https://gateway.infinitescript.com/s/ShapeNetCompletion) and place it under the `data/` directory. The expected directory structure is:
```
│PCN/
├──train/
│  ├── complete
│  │   ├── 02691156
│  │   │   ├── 1a04e3eab45ca15dd86060f189eb133.pcd
│  │   │   ├── .......
│  │   ├── .......
│  ├── partial
│  │   ├── 02691156
│  │   │   ├── 1a04e3eab45ca15dd86060f189eb133
│  │   │   │   ├── 00.pcd
│  │   │   │   ├── 01.pcd
│  │   │   │   ├── .......
│  │   │   │   └── 07.pcd
│  │   │   ├── .......
│  │   ├── .......
├──test/
│  ├── complete
│  │   ├── .......
│  ├── partial
│  │   ├── .......
├──val/
│  ├── complete
│  │   ├── .......
│  ├── partial
│  │   ├── .......
├──PCN.json
└──category.txt
```

## ⚖️ Debiased Evaluation Protocol

To remove pose and scale bias, the centroid and scale are computed **only from the partial input** and applied to both partial and GT:

```python
import numpy as np
from scipy.spatial.transform import Rotation

centroid = pc.mean(0, keepdims=True)
scale = 1.0 / np.max(np.linalg.norm(pc - centroid, axis=1))
R = Rotation.random().as_matrix()

pc = ((pc - centroid) * scale) @ R.T
gt = ((gt - centroid) * scale) @ R.T
```

For evaluation, we recover the original scale before computing metrics:

```python
pred = pred / scale
gt   = gt   / scale
```

## 🚧 TODOs

- [x] Code release
- [x] Debiased evaluation protocol
- [x] HF release

## 🎓 Citation

If you use SIMECO in a scientific work, please consider citing the paper:

<a href="https://arxiv.org/pdf/2509.26631"><img class="image" align="left" width="190px" src="./assets/paper_thumbnail.png"></a>
<a href="https://arxiv.org/pdf/2509.26631">[paper]</a>&nbsp;&nbsp;<a href="https://arxiv.org/abs/2509.26631">[arxiv]</a>&nbsp;&nbsp;<a href="./CITATION.bib">[bibtex]</a><br>
```bibtex
@article{wang2025simeco,
    title={Learning Generalizable Shape Completion with SIM(3) Equivariance}, 
    author={Yuqing Wang and Zhaiyu Chen and Xiao Xiang Zhu},
    journal={arXiv preprint arXiv:2509.26631},
    year={2025}
}
```
<br clear="left"/>

## 🙏 Acknowledgements

Part of our implementation is based on the [PoinTr](https://github.com/yuxumin/PoinTr) repository. We thank the authors for open-sourcing their great work.

simeco

simeco 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

simeco

simeco 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README