---
license: cc-by-4.0
language:
- en
pretty_name: PolypGen2021 Processed Binary Segmentation
task_categories:
- image-to-image
tags:
- medical
- colonoscopy
- polyp-segmentation
- binary-segmentation
size_categories:
- 1K<n<10K
---

# PolypGen2021 Processed Binary Segmentation

Processed PolypGen2021 colonoscopy dataset for prompt-guided binary polyp segmentation.

## Download and extract

Download `PolypGen2021.tar`, then extract it:

```bash
tar -xf PolypGen2021.tar
```

The extracted layout is:

```text
PolypGen2021/
└── PolypGen2021_binary/
    ├── train/
    │   ├── references/
    │   └── targets/
    └── validation/
        ├── references/
        ├── targets/
        └── specs/
```

## Processing summary

- 8,037 total samples.
- Train: 6,429 samples.
- Validation: 1,608 samples.
- All 4,275 source negative samples are retained.
- No test split and no overlay images.
- Reference images are PNG files.
- Binary targets use `[0, 255, 0]` for polyp tissue and `[0, 0, 0]` for background.
- A target is entirely black when no polyp is present.
- Six prompt templates are assigned deterministically with seed 42.
- Validation includes one JSON evaluation spec per sample.

Archive SHA-256:

```text
411f4e3853862b6760f28b694be0bbf9c63149f382519a48962188fb78c9faf2
```

## Source and license

This is a processed derivative of the PolypGen dataset:

- Official project: https://github.com/DebeshJha/PolypGen
- Dataset paper: https://doi.org/10.1038/s41597-023-01981-y
- Source mirror used for this processing run: https://www.kaggle.com/datasets/kokoroou/polypgen2021

The official PolypGen release is published under the Creative Commons Attribution 4.0 International license (CC BY 4.0). Users should cite the original paper and comply with the original dataset terms.

## Citation

```bibtex
@article{ali2023multi,
  title={A multi-centre polyp detection and segmentation dataset for generalisability assessment},
  author={Ali, Sharib and Jha, Debesh and Ghatwary, Noha and Realdon, Stefano and Cannizzaro, Renato and Salem, Osama E and Lamarque, Dominique and Daul, Christian and Riegler, Michael A and Anonsen, Kim V and others},
  journal={Scientific Data},
  volume={10},
  number={1},
  pages={75},
  year={2023}
}
```