https://github.com/boschresearch/Divide-and-Bind

Official implementation of "Divide & Bind Your Attention for Improved Generative Semantic Nursing" (BMVC 2023 Oral)
https://github.com/boschresearch/Divide-and-Bind

aigc diffusion-models generative-model stable-diffusion text-to-image

Last synced: about 2 months ago
JSON representation

Official implementation of "Divide & Bind Your Attention for Improved Generative Semantic Nursing" (BMVC 2023 Oral)

• Host: GitHub
• URL: https://github.com/boschresearch/Divide-and-Bind
• Owner: boschresearch
• License: agpl-3.0
• Created: 2023-09-04T10:27:29.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-01-25T10:08:19.000Z (11 months ago)
• Last Synced: 2024-06-08T02:33:39.678Z (7 months ago)
• Topics: aigc, diffusion-models, generative-model, stable-diffusion, text-to-image
• Language: Jupyter Notebook
• Homepage: https://sites.google.com/view/divide-and-bind
• Size: 17.9 MB
• Stars: 30
• Watchers: 6
• Forks: 3
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-diffusion-categorized - [Code

README

        

# Divide & Bind Your Attention for Improved Generative Semantic Nursing     

:fire:  Official PyTorch implementation for paper 'Divide & Bind Your Attention for Improved Generative Semantic Nursing' (BMVC 2023 **Oral**)

[![arXiv](https://img.shields.io/badge/arXiv-2307.10864-red)](https://arxiv.org/pdf/2307.10864.pdf)    [![Static Badge](https://img.shields.io/badge/Project%20Page-BMVC%20Oral-blue)](https://sites.google.com/view/divide-and-bind)

**:fire: Updates:** We update our [Divide & Bind pipeline](divide_and_bind/pipeline_divide_and_bind_latest.py), which is now compatible with [Diffusers](https://github.com/huggingface/diffusers) 0.21.4!

![overview](docs/overview.png)   

![teaser](docs/teaser.png)   

## Getting Started

Our environment is built on top of [Stable Diffusion](https://github.com/CompVis/stable-diffusion):

```

conda env create -f environment/environment.yaml  

conda activate ldm

```

Additional required packages are listed in [environment/requirement.txt](environment/requirement.txt).    

Stable Diffusion v1.5 model can be found [here](https://huggingface.co/runwayml/stable-diffusion-v1-5).

## Generation

 - [generate_images.ipynb](generate_images.ipynb) provides an example on how to generate images using Divide & Bind.

 -  [examples_generation.ipynb](examples_generation.ipynb) provides some prompts and seeds used in the paper. Example images are saved in [example_outputs](example_outputs).

## Citation

If you use this code please cite

```

@inproceedings{li2023divide,

  title={Divide \& bind your attention for improved generative semantic nursing},

  author={Li, Yumeng and Keuper, Margret and Zhang, Dan and Khoreva, Anna},

  booktitle={34th British Machine Vision Conference 2023, {BMVC} 2023},

  year={2023}

}

```

## License

This project is open-sourced under the AGPL-3.0 license. See the

[LICENSE](LICENSE) file for details.

For a list of other open source components included in this project, see the

file [3rd-party-licenses.txt](3rd-party-licenses.txt).

## Purpose of the project

This software is a research prototype, solely developed for and published as

part of the publication cited above. It will neither be

maintained nor monitored in any way.

## Contact

Please feel free to open an issue or contact personally if you have questions, need help, or need explanations. Don't hesitate to write an email to the following email address:

[email protected]