# MMLA The Wilds: YOLO-Annotated Drone Frames of Painted Dogs, Onagers, Giraffes and Grevy's Zebras

## Facts

- Provider: Imageomics Institute (Kline J.)
- License: CC0-1.0 (https://huggingface.co/datasets/imageomics/mmla_wilds)
- Size: 16,025 files
- Format: JPEG (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/imageomics/mmla_wilds
- Access: public, readable without an account
- Category: Computer Vision
- Source: https://arxiv.org/abs/2504.07744
- Page: https://scigantic.com/archive/267606d9-9259-5436-9d79-78ac2b5d75a0
- Keywords: Computer Vision, JPEG, Hugging Face

## About

Annotated video frames from low-altitude drone flights at The Wilds conservation centre in Ohio, part of MMLA (Multi-Environment, Multi-Species, Low-Altitude Drone Dataset, arXiv 2504.07744). Four sessions in 2024 cover African painted dogs (13,749 frames, DJI Mini), Persian onagers (4,053 frames, Parrot Anafi), giraffes (3,436 frames) and Grevy's zebras (506 frames), 21,744 frames in total at 2720x1530 to 4096x2160. Every frame has a YOLO-format bounding-box label file, annotated by hand in CVAT with kabr-tools. Session folders hold per-video directories of paired .jpg and .txt files, with one video split into partition zips. Measured: 16,025 files, about 55 GB. Companion sets imageomics/mmla_mpala and mmla_opc cover Kenyan sites; together they test whether wildlife detectors generalise across environments, species and drones. A fine-tuned YOLO11m model is published alongside. Dedicated to the public domain under CC0 1.0; cite the dataset and the MMLA paper.

## Load the data

```python
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="imageomics/mmla_wilds", filename="session_1/DJI_0034/DJI_0034_000000.jpg", repo_type="dataset")
print(path)
```

## Data structure

File types in sampled files: JPEG

Example file: `session_1/DJI_0034/DJI_0034_000000.jpg`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/267606d9-9259-5436-9d79-78ac2b5d75a0
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/267606d9-9259-5436-9d79-78ac2b5d75a0
- Over MCP: `get_schema_card` with id `267606d9-9259-5436-9d79-78ac2b5d75a0` at https://api.scigantic.com/mcp

## Related topics

- [JPEG datasets](https://scigantic.com/topics/jpeg)
- [Computer Vision datasets](https://scigantic.com/topics/computer-vision)
