Visual WetlandBirds Dataset

This is the official repository for
Visual WetlandBirds Dataset: Bird Species Identification and Behaviour Recognition in Videos
Javier Rodriguez-Juan, David Ortiz-Perez, Manuel Benavent-Lledo, David Mulero-Pérez, Pablo Ruiz-Ponce, Adrian Orihuela-Torres, Jose Garcia-Rodriguez and Esther Sebastián-González
University of Alicante

Contents

1. Abstract
2. Data
3. Species classification
4. Behavior detection
5. Troubleshooting
6. Citation
7. License
8. Acknowledgements

Abstract

Animal monitoring is a vital field of study, particularly in light of the current biodiversity loss crisis. Data collected through monitoring can provide essential insights and information for decision-making aimed at preserving global biodiversity. Moreover, birds may also respond to human impacts in a more subtle way by modifying their behaviors. Despite the importance of such data, there is a notable scarcity of datasets featuring videos of birds, and no existing dataset offers detailed annotations of bird behaviors in video format. In response to this gap, our study introduces the first fine-grained spatio-temporal dataset specifically designed for bird behavior detection and species classification. This dataset addresses the need for comprehensive bird video datasets and provides detailed data on bird behaviors, facilitating the training of machine learning models to recognize these behaviors, similar to advancements made in human action recognition. The proposed dataset comprises 178 videos recorded in Spanish wetlands, capturing 13 different bird species performing 7 distinct behavior classes.

Data

All videos and annotations can be downloaded from the official Zenodo repository. Make sure you download the latest version uploaded. The following table shows the species, number of videos and minutes recorded for each species:

Common name	Scientific name	Videos	Recorded minutes
Yellow-legged Gull	Larus michahellis	13	5.08
White wagtail	Motacilla alba	13	4.33
Squacco Heron	Ardeola ralloides	15	4.94
Northern shoveler	Spatula clypeata	14	3.49
Mallard	Anas platyrhynchos	10	2.94
Little-ringed plover	Charadrius dubius	10	1.93
Glossy ibis	Plegadis falcinellus	8	3.96
Gadwall	Mareca strepera	13	2.59
Eurasian moorhen	Gallinula chloropus	18	9.18
Eurasian magpie	Pica pica	16	5.95
Eurasian coot	Fulica atra	19	4.11
Black-winged stilt	Himantopus himantopus	14	3.55
Black-headed gull	Chroicocephalus ridibundus	15	6.84

Annotations

Annotation data at frame-level is given for each video. Each frame correspond to one line in the CSV annotations, where multiple lines can be associated with a unique frame. Each line corresponds to the annotation of one bounding box appearing in a specific frame of a video. Each bounding box has the next format:

1. Top-left X-coordinate
2. Top-left Y-coordinate
3. Bottom-right X-coordinate
4. Bottom-right Y-coordinate
5. Behavior ID
6. Subject ID (Identifier of a bird within a video)

Additionally, the species of each bounding box is indicated.

Video samples

final-github-samples.mp4

Species classification

A species classification baseline based on the YOLOv9 model was developed. The results are shown below:

Metric	Result
Precision	0.835
Recall	0.759
mAP50	0.801
mAP50-95	0.556

Behavior detection

A behavior detection baseline based on different video classification models was conducted. The results of the proposed models are presented in the following table:

Model	Learning rate	Accuracy
MViT	0.005	0.51
S3D	0.005	0.29
SwinTransformer	0.009	0.51
ResNet	0.003	0.56

Troubleshooting

For any kind of problems, please read carefully the extended dataset explanation available in the journal paper. We encourage any user to submit any issue related to the data itself using GitHub's built-in issue system within this repository.

Citation

If you use this dataset, please cite:

@article{rodriguez2025visual,
  title={Visual wetlandbirds dataset: Bird species identification and behavior recognition in videos},
  author={Rodriguez-Juan, Javier and Ortiz-Perez, David and Benavent-Lledo, Manuel and Mulero-P{\'e}rez, David and Ruiz-Ponce, Pablo and Orihuela-Torres, Adrian and Garcia-Rodriguez, Jose and Sebasti{\'a}n-Gonz{\'a}lez, Esther},
  journal={Scientific Data},
  volume={12},
  number={1},
  pages={1200},
  year={2025},
  publisher={Nature Publishing Group UK London}
}

@misc{rodriguez2025wetlandbirds,
title={Visual WetlandBirds Dataset: Bird Species Identification and Behaviour Recognition in Videos},
author={Rodriguez-Juan, Javier and Ortiz-Perez, David and Benavent-Lledo, Manuel and Mulero-Pérez, David and Ruiz-Ponce, Pablo and Orihuela-Torres, Adrian and Garcia-Rodriguez, Jose and Sebastián-González, Esther},
month={dec},
year=2024,
publisher={Zenodo},
doi={10.5281/zenodo.14355257},
url={https://doi.org/10.5281/zenodo.14355257}
}

License

The data from the Visual WetlandBirds dataset is released under the MIT license.

Acknowledgements

We would like to thank ”A way of making Europe” European Regional Development Fund (ERDF) and MCIN/AEI/10.13039/501100011033 for supporting this work under the “CHAN-TWIN” project (grant TED2021-130890B- C21 and HORIZON-MSCA-2021-SE-0 action number: 101086387, REMARKABLE, Rural Environmental Monitoring via ultra wide-ARea networKs And distriButed federated Learning. This work is part of the HELEADE project (TSI-100121-2024-24), funded by Spanish Ministry of Digital Processing and by the European Union NextGeneration EU. This work has also been supported by three Spanish national and two regional grants for PhD studies, FPU21/00414, FPU22/04200, FPU23/00532, CIACIF/2021/430 and CIACIF/2022/175.