Rookognise: Acoustic detection and identification of individual rooks in field recordings using multi-task neural networks

Martin, Killian; Adam, Olivier; Obin, Nicolas; Dufour, Valérie

doi:10.1016/j.ecoinf.2022.101818

Cited by 6 publications

(3 citation statements)

References 85 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This manually annotated dataset was used to train a deep neural network. This neural network is described in a previous paper [ 63 ]. In brief, this network replicates the manual annotation process (extracting start and end timestamps and individual identity for each vocal unit) using Mel-scale spectrograms corresponding to short chunks of audio.…”

Section: Methodsmentioning

confidence: 99%

Vocal complexity in a socially complex corvid: gradation, diversity and lack of common call repertoire in male rooks

Martin,

Cornero,

Clayton

et al. 2024

R. Soc. Open Sci.

Self Cite

View full text Add to dashboard Cite

Vocal communication is widespread in animals, with vocal repertoires of varying complexity. The social complexity hypothesis predicts that species may need high vocal complexity to deal with complex social organization (e.g. have a variety of different interindividual relations). We quantified the vocal complexity of two geographically distant captive colonies of rooks, a corvid species with complex social organization and cognitive performances, but understudied vocal abilities. We quantified the diversity and gradation of their repertoire, as well as the inter-individual similarity at the vocal unit level. We found that males produced call units with lower diversity and gradation than females, while song units did not differ between sexes. Surprisingly, while females produced highly similar call repertoires, even between colonies, each individual male produced almost completely different call repertoires from any other individual. These findings question the way male rooks communicate with their social partners. We suggest that each male may actively seek to remain vocally distinct, which could be an asset in their frequently changing social environment. We conclude that inter-individual similarity, an understudied aspect of vocal repertoires, should also be considered as a measure of vocal complexity.

show abstract

Section: Methodsmentioning

confidence: 99%

Vocal complexity in a socially complex corvid: gradation, diversity and lack of common call repertoire in male rooks

Martin,

Cornero,

Clayton

et al. 2024

R. Soc. Open Sci.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Call recognizers are widely used to detect vocal species, but determining the number of calling individuals requires multiple acoustic sensors with source-localization software (Frommolt and Tauchert 2014), because single acoustic sensors cannot provide directionality information (Rone et al 2012). Alternatively, it may be possible, in some cases, to identify individual birds by their idiosyncratic call characteristics using deep learning software (Martin et al 2022).…”

Section: Introductionmentioning

confidence: 99%

A new method to estimate abundance of Australasian Bittern ( Botaurus poiciloptilus ) from acoustic recordings

Znidersic,

Watson,

Towsey

2024

ACE

View full text Add to dashboard Cite

Effective conservation management relies on survey methods that accurately represent the biological communities being monitored. Here, we describe a novel approach using long-duration acoustic recordings to estimate abundance of a threatened wetland bird, the Australasian Bittern (Botaurus poiciloptilus). Whereas acoustic monitoring enables a large increase in effort compared to traditional on-site monitoring, e.g., triangulation surveys, it is difficult to estimate the number of individuals of a target species in acoustic recordings. We describe a semi-automated approach to estimate bittern abundance at four sites in the Barmah-Millewa Forest of southern Australia using single-channel, long-duration recordings. Our approach leveraged several known characteristics of bittern calling behavior. We obtained abundance estimates that are larger than those previously found using triangulation surveys at the same site. This is primarily attributed to our ability to find the peak calling hours in a long-duration recording, which does not require the training of a machine-learning call-recognizer. If the method we describe is performed in a consistent, standardized manner, it can identify population trends, which is an important outcome for a threatened species. Our method should be suitable for other furtive wetland species with a similar call structure or frequency range. Nouvelle méthode pour estimer l'abondance du Butor d'Australie (Botaurus poiciloptilus) à partir d'enregistrementsRÉSUMÉ. Une gestion efficace de la conservation repose sur des méthodes de dénombrement qui représentent les communautés biologiques suivies avec précision. Dans la présente étude, nous décrivons une nouvelle approche au moyen d'enregistrements de longue durée pour estimer l'abondance d'un oiseau menacé de milieux humides, le Butor d'Australie (Botaurus poiciloptilus). Alors que le suivi par enregistrements permet d'augmenter considérablement l'effort par rapport au suivi traditionnel sur site, par exemple les relevés par triangulation, il est difficile d'estimer le nombre d'individus d'une espèce cible dans les enregistrements. Nous présentons une approche semi-automatique pour estimer l'abondance du butor à quatre sites dans la forêt de Barmah-Millewa, dans le sud de l'Australie, au moyen d'enregistrements à canal unique et de longue durée. Notre approche s'appuie sur plusieurs caractéristiques connues du comportement de chant du butor. Nous avons obtenu des estimations d'abondance plus élevées que celles obtenues précédemment à partir de relevés par triangulation sur le même site. Cette différence est principalement attribuable à notre capacité à trouver les heures de pointe des chants dans un enregistrement de longue durée, ne nécessitant pas d'apprentissage automatique pour reconnaître les chants. Si elle est appliquée de manière cohérente et standardisée, notre méthode peut permettre de déterminer la tendance des populations, résultat important pour une espèce menacée. Notre méthode devrait convenir à d'autres espèces furtives de mili...

show abstract

“…This approach has become the main one for bird species recognition and, although large amounts of data are often used, in many cases data augmentation techniques analogous to those used for visual tasks are applied (Kahl et al, 2017). Recently, deep learning techniques started to be applied for the identification of individual birds through their songs (Bedoya and Molles, 2021;Bistel et al, 2022a,b;Martin et al, 2022;Tubaro and Mindlin, 2019). The vocal recognition to track individuals within a population requires that bird call features show lower withinthan between-individual variation and that this variation is stable over the course of an individual's life (Budka et al, 2015).…”

Section: Introductionmentioning

confidence: 99%

Identification of dialects and individuals of globally threatened yellow cardinals using neural networks

Bocaccio,

Domínguez,

Mahler

et al. 2023

Ecological Informatics

View full text Add to dashboard Cite

Rookognise: Acoustic detection and identification of individual rooks in field recordings using multi-task neural networks

Cited by 6 publications

References 85 publications

Vocal complexity in a socially complex corvid: gradation, diversity and lack of common call repertoire in male rooks

Vocal complexity in a socially complex corvid: gradation, diversity and lack of common call repertoire in male rooks

A new method to estimate abundance of Australasian Bittern ( Botaurus poiciloptilus ) from acoustic recordings

Identification of dialects and individuals of globally threatened yellow cardinals using neural networks

Contact Info

Product

Resources

About