An international team of scientists, including researchers from the University of Białystok and the Museum and Institute of Zoology of the Polish Academy of Sciences, has developed the world’s largest global dataset of bird images collected using unmanned aerial vehicles. The open access dataset, called Big Bird, marks a major step forward in the use of artificial intelligence for biodiversity monitoring.
The dataset consists of tens of thousands of drone images gathered across multiple continents and diverse ecosystems. These include seabird colonies, wetlands, coastal zones, agricultural landscapes and inland water bodies. What distinguishes Big Bird from earlier collections is the level of detail in its annotations. Each image has been manually reviewed by experts and labelled not only as containing birds, but also with species identification and, where possible, information on individual positions, group sizes and habitat context.
This level of precision significantly enhances the value of the dataset for training machine learning models. Accurate annotations allow algorithms to learn how to distinguish bird species from aerial imagery with far greater reliability. In practice, this opens the door to automated bird detection, species identification and population counts based on drone surveys. Such tools can dramatically reduce the time, cost and human effort required for large-scale ecological monitoring.
The Big Bird dataset was designed to reflect real-world research conditions. The images vary in flight altitude, camera angle, resolution, lighting conditions and bird density. As a result, machine learning models trained on this dataset are more robust and better suited for application in diverse field scenarios, rather than performing well only under ideal conditions.
The research was published in the journal Remote Sensing in Ecology and Conservation and is expected to become a reference point for future studies combining ecology, remote sensing and artificial intelligence. In the context of accelerating climate change and widespread declines in bird populations, tools that enable faster and more accurate monitoring are becoming increasingly critical.
By contributing to Big Bird, Polish researchers are playing an active role in shaping global standards for data-driven biodiversity research and demonstrating how advanced technologies can support environmental protection at scale.
The original content was published on the website: https://naukawpolsce.pl/aktualnosci/news%2C111625%2Copracowano-najwiekszy-globalny-zbior-danych-zdjec-ptakow-z-dronow.html
