---
license: cc-by-nc-4.0
task_categories:
- object-detection
tags:
- shagai
size_categories:
- n<1K
dataset_info:
  features:
  - name: image
    dtype: image
  - name: objects
    struct:
    - name: bbox
      sequence:
        sequence: int64
    - name: categories
      sequence: int64
  splits:
  - name: train
    num_bytes: 4647508.0
    num_examples: 644
  - name: validation
    num_bytes: 2000239.0
    num_examples: 277
  download_size: 6398185
  dataset_size: 6647747.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
---

# Shagai

Object detection dataset for identifying the shape of [shagai](https://en.wikipedia.org/wiki/Shagai) (ankle of a sheep).

Checkout our original [Github Repo](https://github.com/bilguun0203/ankle-bone-recognition).

## Dataset info

Dataset contains total of 921 images with 3680 objects. Each image has exactly 4 objects.

**Categories:**
- horse
- camel
- sheep
- goat

**Splits:**

- Training: 644 images, 2573 objects, `{0: 644, 1: 642, 2: 668, 3: 619}` (obj. class)
- Validation: 277 images, 1107 objects, `{0: 302, 1: 245, 2: 281, 3: 279}` (obj. class)

## Credits

This dataset was created with contributions from Amarsaikhan Batjargal, Bandikhuu Baasanjav, and Bilguun Ochirbat, students of the National University of Mongolia.

### Citation

```bibtex
@misc{shagai2018,
  author = {Ochirbat, Bilguun and Batjargal, Amarsaikhan and Baasanjav, Bandikhuu},
  title = {{Detect the shape of shagai using RetinaNet}},
  howpublished = {\url{https://github.com/bilguun0203/ankle-bone-recognition}},
  year = {2018},
  month = {July}
}
```