ReCo: Region-Controlled Text-to-Image Generation

by Zhengyuan Yang, Jianfeng Wang, Zhe Gan, Linjie Li, Kevin Lin, Chenfei Wu, Nan Duan, Zicheng Liu, Ce Liu, Michael Zeng, Lijuan Wang

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2023.

******* Update: ReCo is now available at Huggingface Diffusers! COCO Model LAION Model.

Credits to Jaemin Cho. Thank you! *******

ReCo extends T2I models to understand coordinate inputs. Thanks to the introduced position tokens in the region-controlled input query, users can easily specify free-form regional descriptions in arbitrary image regions. For more details, please refer to our paper.

@inproceedings{yang2023reco,
  title={ReCo: Region-Controlled Text-to-Image Generation},
  author={Yang, Zhengyuan and Wang, Jianfeng and Gan, Zhe and Li, Linjie and Lin, Kevin and Wu, Chenfei and Duan, Nan and Liu, Zicheng and Liu, Ce and Zeng, Michael and Wang, Lijuan},
  booktitle={CVPR},
  year={2023}
}

Clone the repository:

git clone https://github.com/microsoft/ReCo.git
cd ReCo

A conda environment named reco_env can be created and activated with:

conda env create -f environment.yaml
conda activate reco_env

Or install packages in requirements.txt:

pip install -r requirements.txt

We recommend using the following AzCopy command to download. AzCopy executable tools can be downloaded here.

Example command:

path/to/azcopy copy <folder-link> <target-address> --resursive"

# For example:
path/to/azcopy copy https://unitab.blob.core.windows.net/data/reco/dataset <local_path> --recursive

Download processed dataset annotations dataset folder in the following dataset path (~59G) with azcopy tool.

path/to/azcopy copy https://unitab.blob.core.windows.net/data/reco/dataset <local_path> --recursive

ReCo checkpoints trained on COCO and a small LAION subset can be downloaded via wget or AzCopy here ReCo_COCO and ReCo_LAION. Save downloaded weights to logs.

inference.sh contains examples for inference calls

eval.sh contains examples for coco evaluation.

For ReCo fine-tuning, we start with the stable diffusion model with instructions here. Weights can be downloaded on HuggingFace. The experiments mainly use sd-v1-4-full-ema.ckpt.

train.sh contains examples for fine-tuning.

The project is built based on the following repository:

• CompVis/stable-diffusion,
• XavierXiao/Dreambooth-Stable-Diffusion,
• harubaru/waifu-diffusion: stable diffusion finetuned on danbooru.

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.

When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact [email protected] with any additional questions or comments.

This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft's Trademark & Brand Guidelines. Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party's policies.