
Wildlife Biology, Journal Year: 2024, Volume and Issue: 2024(6)
Published: Nov. 1, 2024
Language: Английский
Wildlife Biology, Journal Year: 2024, Volume and Issue: 2024(6)
Published: Nov. 1, 2024
Language: Английский
Global Ecology and Conservation, Journal Year: 2024, Volume and Issue: 54, P. e03156 - e03156
Published: Sept. 6, 2024
Language: Английский
Citations
1bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown
Published: April 9, 2024
Abstract 1. Understanding the behavior of animals in their natural habitats is critical to ecology and conservation. Camera traps are a powerful tool collect such data with minimal disturbance. They however produce very large quantity images, which can make human-based annotation cumbersome or even impossible. While automated species identification artificial intelligence has made impressive progress, automatic classification animal behaviors camera trap images remains developing field. 2. Here, we explore potential foundation models, specifically Vision Language Models (VLMs), perform this task without need first train model, would require some level annotation. Using an original dataset alpine fauna annotated by participatory science, investigate zero-shot capabilities different kind recent VLMs predict estimate behavior-specific diel activity patterns three ungulate species. 3. Our results show that using these methods, it possible achieve accuracies over 91% closely align those derived from science (overlap indexes between 84% 90%). 4. These findings demonstrate models vision-language ecological research. Ecologists encouraged adopt new methods leverage full facilitate studies.
Language: Английский
Citations
0Wildlife Biology, Journal Year: 2024, Volume and Issue: 2024(6)
Published: Nov. 1, 2024
Language: Английский
Citations
0