A Mobile Application–Based Citizen Science Product to Compile Bird Observations

Abstract

Citizen science covers initiatives from crowdsourcing, distributed intelligence, and participatory science, to extreme citizen science. Terminological overlap, varied project aims, and cultural differences in the fields of research have, however, led to discord regarding how impactful citizen science projects can be. Here, we showcase a mobile application–based citizen science campaign (in Finnish: Muuttolintujen kevät), an automated bird sound classifier of Finnish birds. Over a single season (2023), the method attracted 140,000 participants who uploaded close to three million recordings containing six million bird observations. We report the spatial and temporal distribution of the observations collected, characterize the user behaviour, and discuss reliability of the user-based validations of the AI-powered species identifications. To circumvent data quality problems that characterize many citizen science projects, our approach stores the raw audio in a centralized repository, enabling rigorous validation and re-analysis. Mobile application-based citizen science initiatives can be harnessed to probe the state of our environment almost in real time and potentially guide conservation acts in the future.