Skip to content

Official implementation of the paper "RoomEditor: High-Fidelity Furniture Synthesis with Parameter-Sharing U-Net"

Notifications You must be signed in to change notification settings

stonecutter-21/roomeditor

Repository files navigation

[NeurIPS 2025] RoomEditor: High-Fidelity Furniture Synthesis with Parameter-Sharing U-Net

alt text

🔥 Updates

[2025/12/22] Release the RoomEditor++ model, the RoomBench++ dataset. PDF

[2025/12/15] Release the RoomEditor model, the RoomBench dataset, and the evaluation code.

Requirements

To install requirements:

pip install -r requirements.txt

To install RoomEditor++ requirements:

pip install -r RoomEditor++/requirements.txt

Download Checkpoints

If you need the all train data, please sent email to contact us.

Download SD-1.5-inpainting checkpoint:

Download Roomeditor:

  • You could download the model weights and test images from here

Download Roomeditor++:

  • You could download the model weights and test images from here

Evaluation

Run inference with

bash run.sh 

Run RoomEditor++ inference with

python /root/autodl-fs/code/RoomEditor/RoomEditor++/inference.py

Gradio Demo

You can run this script:

python run_gradio_demo.py

Citation

@inproceedings{
lin2025roomeditor,
title={RoomEditor: High-Fidelity Furniture Synthesis with Parameter-Sharing U-Net},
author={Zhenyi Lin and Xiaofan Ming and Qilong Wang and Dongwei Ren and Wangmeng Zuo and Qinghua Hu},
booktitle={Advances in Neural Information Processing Systems},
year={2025}
}

About

Official implementation of the paper "RoomEditor: High-Fidelity Furniture Synthesis with Parameter-Sharing U-Net"

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published