---
license: cc-by-4.0
task_categories:
- object-detection
pretty_name: DonaDataset
---

# DonaDataset

![License](https://img.shields.io/badge/license-CC--BY--4.0-blue.svg)
[![WildINTEL](https://img.shields.io/badge/WildINTEL-project-blue)](https://wildintel.eu/)

Version: `REPLACE_WITH_VERSION`

Camera-trap image dataset for object detection, from Doñana National Park, Spain.

## Dataset format

This dataset is distributed in YOLO format for object detection.

After downloading and extracting all tar shards into the same directory, the expected structure is:

```text
.
├── donana.yaml
├── images/
│   ├── train/
│   ├── val/
│   └── test/
└── labels/
    ├── train/
    ├── val/
    └── test/
```

The YOLO configuration file uses relative paths:

```yaml
path: .
train: images/train
val: images/val
test: images/test
```

## Splits

- `train`: 3 images
- `val`: 3 images
- `test`: 3 images

## Classes

- `0`: Ave
- `1`: Felis catus
- `2`: Bos taurus
- `3`: Canis familiaris
- `4`: Herpestes ichneumon
- `5`: Meles meles
- `6`: Oryctolagus cuniculus
- `7`: Dama dama
- `8`: Genetta genetta
- `9`: Equus sp
- `10`: Lepus granatensis
- `11`: Lynx pardinus
- `12`: Cervus elaphus
- `13`: Vulpes vulpes
- `14`: Sus scrofa
- `15`: Empty

## Data files

The image and label files are stored inside tar shards under:

```text
data/train/
data/val/
data/test/
```

Each tar file preserves YOLO-compatible internal paths, for example:

```text
images/train/example.jpg
labels/train/example.txt
```

## Integrity verification

The export includes:

```text
checksums-sha256.txt
manifest.csv
manifest-files-sha256.csv
verification_report_local.json
```

Use these files to verify downloaded shards and internal image/label hashes.

## Example extraction

From the directory containing the downloaded files:

```bash
mkdir -p extracted
cp donana.yaml extracted/

for f in data/train/*.tar data/val/*.tar data/test/*.tar; do
  tar -xf "$f" -C extracted/
done
```

Then train with a YOLO-compatible framework using:

```text
extracted/donana.yaml
```

## Hugging Face repository

```text
wildintelproject/donadataset
```

## Contributing

Issues and pull requests are welcome in the source code repository: https://github.com/wildintelproject/donadataset

## Funding

This work is part of the [WildINTEL project](https://wildintel.eu/), funded by the
[Biodiversa+](https://www.biodiversa.eu/) Joint Research Call 2022-2023
"Improved transnational monitoring of biodiversity and ecosystem change for science and society (BiodivMon)".

## License

This dataset is released under: `CC-BY-4.0`.

## Citation

Please check `CITATION.cff` for citation information.

## Limitations and ethical considerations

Before using this dataset, users should review possible limitations related to camera-trap data, class imbalance, annotation quality, environmental bias, privacy, and sensitive ecological information.