Dataset of synthetic colonoscopy images, generated with Deambooth fine-tuned Stable Diffusion, for training polyp detection models.

This dataset consists of 253 synthetic colorectal polyp images generated using a Dreambooth fine- tuned Stable Diffusion v1.5 pipeline, together with manually annotated bounding boxes for object detection tasks.

Dataset description

This dataset is provided for research and educational purposes only. It is not intended for direct clinical use, medical decision-making, or commercial deployment without further validation. <br></br>This release contains synthetic polyp images and associated bounding-box annotations only.<br></br>Users of this dataset are requested to cite the following publication:<br></br><blockquote>Chng, S., Li, J., Trinh, H., Lim, J., Kan, M., Vadapalli, D., Toh, P., & Ang, T. (2026). Polyp Detection Using YOLOv9 on Real and Synthetic Colonoscopy Images. <i>Journal of Gastroenterology and Hepatology.</i> https://doi.org/10.1111/jgh.70391<br></br></blockquote>in any academic work, publications, or derivative works that make use of this dataset.

Terms of use

Examples

<a href='/assets/datasets/synthetic-polyps.zip' download> synthetic-polyps.zip </a>