An open API service indexing awesome lists of open source software.

https://github.com/naver-ai/densediffusion

Official Pytorch Implementation of DenseDiffusion (ICCV 2023)
https://github.com/naver-ai/densediffusion

Last synced: 7 months ago
JSON representation

Official Pytorch Implementation of DenseDiffusion (ICCV 2023)

Awesome Lists containing this project

README

          

## Dense Text-to-Image Generation with Attention Modulation
### ICCV 2023 [[Paper](https://arxiv.org/abs/2308.12964)] [[Demo on HF 🤗](https://huggingface.co/spaces/naver-ai/DenseDiffusion)] [[Colab Demo](https://github.com/XandrChris/DenseDiffusionColab)]


> #### Authors    [Yunji Kim](https://github.com/YunjiKim)1, [Jiyoung Lee](https://lee-jiyoung.github.io)1, [Jin-Hwa Kim](http://wityworks.com/)1, [Jung-Woo Ha](https://github.com/jungwoo-ha)1, [Jun-Yan Zhu](https://www.cs.cmu.edu/~junyanz/)2
         1NAVER AI Lab, 2Carnegie Mellon University

> #### Abstract
Existing text-to-image diffusion models struggle to synthesize realistic images given dense captions, where each text prompt provides a detailed description for a specific image region.
To address this, we propose DenseDiffusion, a training-free method that adapts a pre-trained text-to-image model to handle such dense captions while offering control over the scene layout.
We first analyze the relationship between generated images' layouts and the pre-trained model's intermediate attention maps.
Next, we develop an attention modulation method that guides objects to appear in specific regions according to layout guidance.
Without requiring additional fine-tuning or datasets, we improve image generation performance given dense captions regarding both automatic and human evaluation scores.
In addition, we achieve similar-quality visual results with models specifically trained with layout conditions.

> #### Method






Our goal is to improve the text-to-image model's ability to reflect textual and spatial conditions without fine-tuning.
We formally define our condition as a set of $N$ segments ${\lbrace(c_{n},m_{n})\rbrace}^{N}_{n=1}$, where each segment $(c_n,m_n)$ describes a single region.
Here $c_n$ is a non-overlapping part of the full-text caption $c$, and $m_n$ denotes a binary map representing each region. Given the input conditions, we modulate attention maps of all attention layers on the fly so that the object described by $c_n$ can be generated in the corresponding region $m_n$.
To maintain the pre-trained model's generation capacity, we design the modulation to consider original value range and each segment's area.

> #### Examples







----

### How to launch a web interface

- Put your access token to Hugging Face Hub [here](./gradio_app.py#L77).

- Run the Gradio app.
```
python gradio_app.py
```

----

### Getting Started

- Create the image layout.



- Label each segment with a text prompt.



- Adjust the full text. The default full text is automatically concatenated from each segment's text. The default one works well, but refineing the full text will further improve the result.



- Check the generated images, and tune the hyperparameters if needed.

wc : The degree of attention modulation at cross-attention layers.

ws : The degree of attention modulation at self-attention layers.



----

### Benchmark

We share the benchmark used in our model development and evaluation [here](./dataset).
The code for preprocessing segment conditions is in [here](./inference.ipynb).

---

#### BibTeX
```
@inproceedings{densediffusion,
title={Dense Text-to-Image Generation with Attention Modulation},
author={Kim, Yunji and Lee, Jiyoung and Kim, Jin-Hwa and Ha, Jung-Woo and Zhu, Jun-Yan},
year={2023},
booktitle = {ICCV}
}
```

---

#### Acknowledgment
The demo was developed referencing this [source code](https://huggingface.co/spaces/weizmannscience/multidiffusion-region-based). Thanks for the inspiring work! 🙏