README.md

---
license: apache-2.0
pipeline_tag: image-to-3d
---

# Lite3R: A Model-Agnostic Framework for Efficient Feed-Forward 3D Reconstruction

Official implementation of **Lite3R**, a model-agnostic framework for efficient feed-forward 3D reconstruction from multi-view images.

Lite3R introduces a systematic approach to compress large-scale 3D reconstruction models while maintaining reconstruction quality. The framework combines Sparse Linear Attention (SLA), FP8-Aware Quantization-Aware Training (QAT), and Partial Attention Distillation.

[**Paper**](https://huggingface.co/papers/2605.11354) | [**Website**](https://aigeeksgroup.github.io/Lite3R/) | [**Code**](https://github.com/AIGeeksGroup/Lite3R)

## Installation

```bash
# Clone the repository
git clone https://github.com/AIGeeksGroup/Lite3R.git
cd Lite3R

# Create conda environment
conda create -n lite3r python=3.10
conda activate lite3r

# Install dependencies
pip install -r requirements.txt
```

## Sample Usage

### Inference

After downloading the model checkpoints from this repository, you can run inference using the following command:

```bash
python inference.py \
  --model vggt \
  --checkpoint checkpoints/fp8_qat_1ep/vggt/vggt_fp8_qat_1ep.pt \
  --input_dir examples/input \
  --output_dir examples/output
```

## Citation

If you find this work useful, please cite:

```bibtex
@article{zhang2026lite3r,
  title={Lite3R: A Model-Agnostic Framework for Efficient Feed-Forward 3D Reconstruction},
  author={Zhang, Haoyu and Zhang, Zeyu and Zhou, Zedong and Zhao, Yang and Tang, Hao},
  journal={arXiv preprint arXiv:2605.11354},
  year={2026}
}
```