Assessing topographic features and population abundance in an Antarctic penguin colony through UAV-based deep-learning models
Penguins play an essential biochemical role in the Antarctic ecosystem, being the study of their dynamics of utmost importance to understand their environment, behaviour and populational trends in the current climate change scenario. In this study, we used multi-rotor Unmanned Aerial Vehicles (UAVs)...
| Autores: | , , , , , |
|---|---|
| Formato: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2024 |
| País: | España |
| Recursos: | Consejo Superior de Investigaciones Científicas (CSIC) |
| Repositorio: | DIGITAL.CSIC. Repositorio Institucional del CSIC |
| OAI Identifier: | oai:digital.csic.es:10261/378279 |
| Acesso em linha: | http://hdl.handle.net/10261/378279 https://api.elsevier.com/content/abstract/scopus_id/85203536787 |
| Access Level: | acceso abierto |
| Palavra-chave: | UAV Deep-learning Chinstrap penguin Pygoscelis antarcticus Breeding site features Vapour col |
| Resumo: | Penguins play an essential biochemical role in the Antarctic ecosystem, being the study of their dynamics of utmost importance to understand their environment, behaviour and populational trends in the current climate change scenario. In this study, we used multi-rotor Unmanned Aerial Vehicles (UAVs) along the coast of the chinstrap penguin (Pygoscelis antarcticus) colony of Vapour Col (Deception Island, Antarctica) to map potential sites of biochemical interactions with the surrounding sea water. Several runoff discharge points were identified, where a precise placing of environmental sampling station is suggested. Additionally, UAVs were used in combination with Object Detection Architectures to obtain the chinstrap colony population size. Applying a simulation for clutch initiation dates due to our off-laying peak count, we obtained an estimated range of 13,250 to 22,000 breeding pairs in the 2021/2022 breeding season, also suggesting an alternative approach using chinstrap chicks as proxy to estimate adult numbers. This research shows the utility of UAV-deep learning for environment characterization and wildlife monitoring, providing a solid framework for upcoming studies in the area. |
|---|