README.md

---
license: cc0-1.0
language:
- en
pretty_name: MMLA Ol Pejeta Conservancy
task_categories:
- image-classification
tags:
- biology
- image
- animals
- CV
- drone
- zebra
size_categories: 10K<n<100K
description: "Annotated video frames of Plains zebras collected at the Ol Pejeta Conservancy (OPC) in Kenya using the semi-autonomous WildWing system. includes frames from various sessions, with annotations indicating the presence of zebras in the images in YOLO format. The dataset is intended for use in training and evaluating computer vision models for animal detection and classification from drone imagery to facilitate research in wildlife monitoring and conservation using advanced imaging technologies."
---


# Dataset Card for MMLA Ol Pejeta Conservancy

<!-- Provide a quick summary of what the dataset is or can be used for. --> 

## Dataset Details
This is a dataset containing annotated video frames of Plains zebras collected at the [Ol Pejeta Conservancy (OPC)](https://www.olpejetaconservancy.org/) in Kenya using the semi-autonomous [WildWing system](https://imageomics.github.io/wildwing/). The dataset is intended for use in training and evaluating computer vision models for animal detection and classification from drone imagery. It includes frames from various sessions, 
with annotations indicating the presence of zebras in the images in YOLO format, and is designed to facilitate research in wildlife monitoring and conservation using advanced imaging technologies.


### Dataset Description


- **Curated by:** Jenna Kline
- **Homepage:** [MMLA project](https://github.com/Imageomics/mmla)
- **Repository:** [Imageomics/mmla](https://github.com/Imageomics/mmla)
- **Paper:** [MMLA: Multi-Environment, Multi-Species, Low-Altitude Aerial Footage Dataset](https://arxiv.org/abs/2504.07744)


<!-- Provide a longer summary of what this dataset is. -->
This dataset contains video frames collected using the [WildWing system](https://imageomics.github.io/wildwing/), which is an semi-autonomous drone designed for wildlife monitoring. 

The dataset includes frames from multiple sessions, over two days of data collection, 2025-01-31 and 2025-02-01, with a total of 5 videos. Each session captures video footage of Plains zebras in their natural habitat at the Ol Pejeta Conservancy in Kenya. 

The dataset consists of 29,268 frames. Each frame is accompanied by annotations in YOLO format, indicating the presence of zebras and their bounding boxes within the images. The annotations were completed manually by the dataset curator using [CVAT](https://www.cvat.ai/) and [kabr-tools](https://github.com/Imageomics/kabr-tools).


The dataset is intended for use in training and evaluating computer vision models for animal detection and classification from drone imagery. 

See the [fine-tuned YOLO11m model](https://huggingface.co/imageomics/mmla) that was trained using this dataset.



| Session | Date Collected | Video ID  | Total Frames | Size (pixels) | 
|---------|---------------|-----------|--------------|---------------|
| `session_1`       | 2025-01-31    | P0800081  | 5,949        | 3840x2160     |
| `session_1`       | 2025-01-31    | P0830086  | 2,439        | 3840x2160     |
| `session_1`       | 2025-01-31    | P0840087  | 4,461        | 4096x2160     |
| `session_1`       | 2025-01-31    | P0860090  | 1,754        | 3840x2160     |
| `session_1`       | 2025-01-31    | P0870091  | 2,123        | 4096x2160     |
| `session_2`       | 2025-02-01    | P0910095  | 5,978        | 4096x2160     |
| `session_2`       | 2025-02-01    | P0940098  | 6,564        | 4096x2160     |
| **Total Frames:** |  |  | **29,268** |  |

This table shows the data collected at Ol Pejeta Conservancy in Laikipia, Kenya, with session information, dates, frame counts, and pixel resolution.

The dataset includes frames extracted from drone videos captured during five distinct data collection sessions. Each session represents a separate field excursion lasting approximately one hour, conducted at a specific geographic location. 
Multiple sessions may occur on the same day but in different locations or targeting different animal groups. During each session, multiple drone videos were recorded to capture animals in their natural habitat under varying environmental conditions.

## Dataset Structure
```
/dataset/
    classes.txt
    session_1/
        P0800081/
            partition_1/
                P0800081_000000.jpg
                P0800081_000000.txt
                ...
                P0800081_007099.txt
            partition_2/
                P0800081_007100.jpg
                P0800081_007100.txt
                ...
                P0800081_010048.txt
        P0830086/
            P0830086_000000.jpg
            P0830086_000000.txt
            ...
            P0830086_002438.txt
        P0840087/
            P0840087_000000.jpg
            P0840087_000000.txt
            ...
            P0840087_004770.txt
        P0860090/
            P0860090_000000.jpg
            P0860090_000000.txt
            ...
            P0860090_001753.txt
        P0870091/
            P0870091_20250311_000000.jpg
            P0870091_20250311_000000.txt
            ...
            P0870091_20250311_003060.txt
    session_2/
        P0910095/
            partition_1/
                P0910095_000000.jpg
                P0910095_000000.txt
                ...
                P0910095_002999.txt
            partition_2/
                P0910095_003000.jpg
                P0910095_003000.txt
                ...
                P0910095_005977.txt
        P0940098/
            partition_1/
                P0940098_20250311_000000.jpg
                P0940098_20250311_000000.txt
                ...
                P0940098_20250311_003499.txt
            partition_2/
                P0940098_20250311_003500.jpg
                P0940098_20250311_003500.txt
                ...
                P0940098_20250311_006563.txt
```

### Data Instances
All images are named `<video_id>_<frame_number>.jpg`, under the particular session and full video to which they belong; these can be matched to dates based on the table above. The annotations are in YOLO format and are stored in a corresponding `.txt` file with the same name as the image. 2025-01-31 and 2025-02-01 are the two days of data collection, with a total of 7 sessions. 2025-01-31 has 5 sessions and 2025-02-01 has 2 sessions. 

Note on data partitions: Hugging Face limits folders to 10,000 files per folder, so each video file is further divided into partitions of 10,000 files. The partition folders are named `partition_1`, `partition_2`, etc.


### Data Fields

**classes.txt**:
  - `0`: zebra
  - `1`: giraffe
  - `2`: onager
  - `3`: dog

Note: only zebras appear in this dataset; other class labels are included to be consistent across MMLA data collected at other locations, 
see the [MMLA data from Mpala Research Center](https://huggingface.co/datasets/imageomics/mmla_mpala) and [The Wilds MMLA dataset](https://huggingface.co/datasets/imageomics/mmla_wilds).


**frame_id.txt**:
  - `class`: Class of the object in the image (0 for zebra)
  - `x_center`: X coordinate of the center of the bounding box (normalized to [0, 1])
  - `y_center`: Y coordinate of the center of the bounding box (normalized to [0, 1])
  - `width`: Width of the bounding box (normalized to [0, 1])
  - `height`: Height of the bounding box (normalized to [0, 1])

### Data Splits

This dataset was used in conjunction with the other two [MMLA datasets](https://huggingface.co/collections/imageomics/mmla) for both training and testing the [MMLA YOLO model](https://huggingface.co/imageomics/mmla#training-details).


## Dataset Creation

### Curation Rationale

<!-- Motivation for the creation of this dataset. For instance, what you intended to study and why that required curation of a new dataset (or if it's newly collected data and why the data was collected (intended use)), etc. -->

The dataset was created to facilitate research in wildlife monitoring and conservation using advanced imaging technologies. The goal is to develop and evaluate computer vision models that can accurately detect and classify animals from drone imagery, and their generalizability across different species and environments.


<!-- ### Source Data -->

<!-- This section describes the source data (e.g., news text and headlines, social media posts, translated sentences, ...). As well as an original source it was created from (e.g., sampling from Zenodo records, compiling images from different aggregators, etc.) -->

#### Data Collection and Processing

<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, re-sizing of images, tools and libraries used, etc. 
This is what _you_ did to it following collection from the original source; it will be overall processing if you collected the data initially.
-->

The data was collected using the [WildWing](https://github.com/Imageomics/wildwing) system, which semi-autonomously captures video footage of wildlife in their natural habitat. The data collection process involved flying the drone over the [Ol Pejeta Conservancy](https://www.olpejetaconservancy.org/) in Kenya, where Plains zebras were observed. The missions were flown during the [WildDrone](https://wilddrone.eu/) Hackathon in January 2025, with the goal of capturing high-quality video footage for ecological analysis.

The videos were annotated manually using the Computer Vision Annotation Tool [CVAT](https://www.cvat.ai/) and [kabr-tools](https://github.com/Imageomics/kabr-tools) library. These detection annotations and original video files were then processed to extract individual frames, which were saved as JPEG images. The annotations were converted to YOLO format, with bounding boxes indicating the presence of zebras in each frame.

<!-- #### Who are the source data producers?
[More Information Needed] -->
<!-- This section describes the people or systems who originally created the data.

Ex: This dataset is a collection of images taken of the butterfly collection housed at the Ohio State University Museum of Biological Diversity. The associated labels and metadata are the information provided with the collection from biologists that study butterflies and supplied the specimens to the museum.
 -->


### Annotations
<!-- 
If the dataset contains annotations which are not part of the initial data collection, use this section to describe them. 

Ex: We standardized the taxonomic labels provided by the various data sources to conform to a uniform 7-rank Linnean structure. (Then, under annotation process, describe how this was done: Our sources used different names for the same kingdom (both _Animalia_ and _Metazoa_), so we chose one for all (_Animalia_). -->

#### Annotation process
CVAT and [kabr-tools](https://github.com/Imageomics/kabr-tools) were used to annotate the video frames. The annotation process involved manually labeling the presence of zebras in each frame, drawing bounding boxes around them, and converting the annotations to YOLO format.
<!-- This section describes the annotation process such as annotation tools used, the amount of data annotated, annotation guidelines provided to the annotators, interannotator statistics, annotation validation, etc. -->

#### Who are the annotators?
Jenna Kline
<!-- This section describes the people or systems who created the annotations. -->

### Personal and Sensitive Information
The dataset was cleaned to remove any personal or sensitive information. All images are of Plains zebras in their natural habitat, and no identifiable human subjects are present in the dataset.
<!-- 
For instance, if your data includes people or endangered species. -->


<!-- ## Considerations for Using the Data
[More Information Needed]
<!--
Things to consider while working with the dataset. For instance, maybe there are hybrids and they are labeled in the `hybrid_stat` column, so to get a subset without hybrids, subset to all instances in the metadata file such that `hybrid_stat` is _not_ "hybrid".
-->

<!-- ### Bias, Risks, and Limitations
[More Information Needed] -->
<!-- This section is meant to convey both technical and sociotechnical limitations. Could also address misuse, malicious use, and uses that the dataset will not work well for.-->

<!-- For instance, if your data exhibits a long-tailed distribution (and why). -->

<!-- ### Recommendations
[More Information Needed]
This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->

## Licensing Information

This dataset has been marked as dedicated to the public domain by applying the [CC0-1.0 Public Domain Waiver](https://creativecommons.org/publicdomain/zero/1.0/). We ask that you cite the dataset and paper using the below citations if you make use of it in your research.


## Citation

**BibTeX:**

**Data**
```
@misc{mmla_opc,
  author = {Kline, Jenna and 
            Nguyen Ngoc, Dat and
            Hine, Duncan and 
            Rondeau Saint-Jean, Camille and
            Maalouf, Guy and
            Juma, Brenda and
            Kilwaya, Alex and
            Vuyiya, Brian and
            Macharia, Irungu and
            Njoroge, William and
            Mutisya, Samuel and
            Guerin, David and
            Costelloe, Blair and
            Pastucha, Elzbieta and
            Hermansen, Jussi and
            Jensen, Kjeld and
            Watson, Matt and
            Richardson, Tom and
            Pagh Schultz Lundquist, Ulrik  
          },
  title = {MMLA Ol Pejeta Conservancy (OPC) Dataset (Revision e81c0d9)},
  year = {2025},
  url = {https://huggingface.co/datasets/imageomics/mmla_opc},
  doi = {10.57967/hf/7378},
  publisher = {Hugging Face}
}
```

**Paper**
```
@misc{kline2025mmla,
      title={MMLA: Multi-Environment, Multi-Species, Low-Altitude Drone Dataset}, 
      author={Jenna Kline and Samuel Stevens and Guy Maalouf and Camille Rondeau Saint-Jean and Dat Nguyen Ngoc and Majid Mirmehdi and David Guerin and Tilo Burghardt and Elzbieta Pastucha and Blair Costelloe and Matthew Watson and Thomas Richardson and Ulrik Pagh Schultz Lundquist},
      year={2025},
      eprint={2504.07744},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2504.07744}, 
}
```


## Acknowledgements

This work was supported by the [WildDroneEU Project](https://wilddrone.eu). WildDrone is an MSCA Doctoral Network funded by the European Union’s Horizon Europe research and innovation funding programme under the Marie Skłodowska-Curie grant agreement no. 101071224.

This work was supported by the [Imageomics Institute](https://imageomics.org), which is funded by the US National Science Foundation's Harnessing the Data Revolution (HDR) program under [Award #2118240](https://www.nsf.gov/awardsearch/showAward?AWD_ID=2118240) (Imageomics: A New Frontier of Biological Information Powered by Knowledge-Guided Machine Learning). 

This work was supported by the AI Institute for Intelligent Cyberinfrastructure with Computational Learning in the Environment [ICICLE](https://icicle.osu.edu/), which is funded by the US National Science Foundation under grant number OAC-2112606.

Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

<!-- You may also want to credit the source of your data, i.e., if you went to a museum or nature preserve to collect it. -->

<!-- ## Glossary  -->

<!-- [optional] If relevant, include terms and calculations in this section that can help readers understand the dataset or dataset card. -->

## More Information 
The data was collected under Kenya Civil Aviation Authority (KCAA) permit number KCAA/UAS/OPS/0048/2025. The data collection was conducted in collaboration with the Ol Pejeta Conservancy and the WildDrone Hackathon team in accordance with Research License No. NACOSTI/P/25/415376. 

<!-- [optional] Any other relevant information that doesn't fit elsewhere. -->

## Dataset Card Authors 

Jenna Kline

## Dataset Card Contact

kline.377 at osu.edu
<!-- Could include who to contact with questions, but this is also what the "Discussions" tab is for. -->