README.md

---
task_categories:
- image-segmentation
size_categories:
- 1K<n<10K
tags:
- not-for-all-audiences
---
# Danbooru SFW 512px Character Crops
This dataset includes image and bounding box pairs where the images are samples from 
Danbooru 2021 SFW and the bounding boxes are crops of the central character(s). It can 
be used to train a model that automatically crops the base dataset in order to create a cleaner
and more character-centric version. Importantly, the crops aim to remove artifacts such as
horizontal/vertical bars that could negatively impact performance on image generation tasks.

## Bounding Boxes
The base images are all 512x512, the bounding boxes represent crops with a maximium size of
512x512 and a minimum size of 128x128. Each bounding box is represented as a 3-dimensional 
vector [x, y, s] where x is the left coordinate of the box, y is the top coordinate of the box,
and s is the side length of the box. These elements have been converted to floats and 
normalized according to their maximum possible values (512). For example, [16, 28, 256] would 
have been converted to [16/512, 28/512, 256/512], or [0.03125, 0.0546875, 0.5]. Using this 
conversion, the minimum value of s is 128/512, or 0.25.

## Original Dataset Citation
```bibtex
@misc{danbooru2021,
    author={Anonymous and Danbooru community and Gwern Branwen},
    title={Danbooru2021: A Large-Scale Crowdsourced and Tagged Anime Illustration Dataset},
    howpublished={\url{https://gwern.net/danbooru2021}},
    url={https://gwern.net/danbooru2021},
    type={dataset},
    year={2022},
    month={January},
    timestamp={2022-01-21},
    note={Accessed: 2023-12-06}
}
```