Automated detection of gibbon calls from passive acoustic monitoring data using convolutional neural networks in the "torch for R" ecosystem

TL;DR

This study develops and benchmarks convolutional neural network models within the R environment for automated detection of gibbon calls from passive acoustic monitoring data, enabling ecological research and conservation efforts.

Contribution

It introduces a native R-based deep learning workflow for gibbon call detection, compares six CNN architectures, and applies the best models to real-world ecological datasets.

Findings

Model performance varies by species and dataset.

Top models successfully detect gibbon calls in field data.

Provides benchmarks for future acoustic monitoring research.

Abstract

Automated detection of acoustic signals is crucial for effective monitoring of sound-producing animals and their habitats across ecologically relevant spatial and temporal scales. Recent advances in deep learning have made these approaches more accessible. However, few deep learning approaches can be implemented natively in the R programming environment; approaches that run natively in R may be more accessible for ecologists. The "torch for R" ecosystem has made deep learning with convolutional neural networks accessible for R users. Here, we evaluate a workflow for the automated detection and classification of acoustic signals from passive acoustic monitoring (PAM) data. Our specific goals include: 1) present a method for automated detection of gibbon calls from PAM data using the "torch for R" ecosystem; 2) conduct a series of benchmarking experiments and compare the results of six CNN architectures; and 3) investigate how well the different architectures perform on datasets of the female calls from two different gibbon species: the northern grey gibbon (Hylobates funereus) and the southern yellow-cheeked crested gibbon (Nomascus gabriellae). We found that the highest-performing architecture depended on the species and test dataset. We successfully deployed the top-performing model for each gibbon species to investigate spatial variation in gibbon calling behavior across two grids of autonomous recording units in Danum Valley Conservation Area, Malaysia and Keo Seima Wildlife Sanctuary, Cambodia. The fields of deep learning and automated detection are rapidly evolving, and we provide the methods and datasets as benchmarks for future work.