---
license: mit
language:
- en
pretty_name: Polyp-Gen
---


# Polyp-Gen: Realistic and Diverse Polyp Image Generation for Endoscopic Dataset Expansion

Polyp-Gen is a text-guided full-automatic diffusion-based endoscopic image generation framework for realistic and diverse polyp image generation for endoscopic dataset expansion, as presented in [Polyp-Gen: Realistic and Diverse Polyp Image Generation for Endoscopic Dataset Expansion](https://huggingface.co/papers/2501.16679). You can use our model for polyp generation.

Code is available [here](https://github.com/CUHK-AIM-Group/Polyp-Gen).

## Dataset

This dataset was modified by the original [LDPolypVideo](https://github.com/dashishi/LDPolypVideo-Benchmark) dataset.

We filtered out some low-quality images with blurry, reflective, and ghosting effects, and finally select 55,883 samples including 29,640 polyp frames and 26,243 non-polyp frames. 

[02/26] We update the download link of the training and test dataset at HuggingFace [link](https://huggingface.co/datasets/Saint-lsy/Polyp-Gen-Dataset)


## Citation

If you find this work helpful, please consider to **star🌟** this repo and cite the following paper:

```bib
@article{liu2025polyp,
  title={Polyp-Gen: Realistic and Diverse Polyp Image Generation for Endoscopic Dataset Expansion},
  author={Liu, Shengyuan and Chen, Zhen and Yang, Qiushi and Yu, Weihao and Dong, Di and Hu, Jiancong and Yuan, Yixuan},
  journal={arXiv preprint arXiv:2501.16679},
  year={2025}
}
```
and the original LDPolypVideo paper:
```
Yiting. Ma, Xuejin. Chen, Kai. Cheng, Yang. Li and Bin. Sun. "LDPolypVideo Benchmark: A Large-scale Colonoscopy Video Dataset of Diverse Polyps", Medical Image Computing and Computer Assisted Intervention Society, 2021
```