Szczegóły publikacji
Opis bibliograficzny
Deep learning methods for acoustic monitoring of birds migrating at night / Hanna PAMUŁA, Agnieszka Pocha, Maciej KŁACZYŃSKI // W: e-Forum acusticum 2020 [Dokument elektroniczny] : 7–11 December 2020, Lyon, France. — Wersja do Windows. — Dane tekstowe. — [Lyon : INSAVALOR], [2020]. — S. 2761–2764. — Wymagania systemowe: Adobe Reader. — Tryb dostępu: https://hal.archives-ouvertes.fr/hal-03242466/document [2022-04-27]. — Bibliogr. s. 2764, Abstr.
Autorzy (3)
- AGHPamuła Hanna
- Pocha Agnieszka
- AGHKłaczyński Maciej Krzysztof
Dane bibliometryczne
| ID BaDAP | 139988 |
|---|---|
| Data dodania do BaDAP | 2022-04-27 |
| DOI | 10.48465/fa.2020.0650 |
| Rok publikacji | 2020 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
Abstract
The project aims to supplement nocturnal bird migration research with long-term audio recordings. We checked how different deep learning methods – namely convolutional neural networks (CNN) and residual neural networks (ResNets) – perform in the nocturnal flight calls detection task. Moreover, we used transfer learning from image classification models to determine if we can improve automatic detection performance. The best model obtained the AUC score over 95% (area under the receiver operating characteristic curve). Although higher bird detection accuracy is still needed because of the strong dataset imbalance, the results are very promising and suggest that deep learning models applied to audio data have great potential in supplementing bird migration studies.
