Towards Deep Active Learning in Avian Bioacoustics

TL;DR

This paper explores deep active learning methods to improve avian bioacoustics monitoring by reducing annotation efforts and enhancing model adaptability across diverse environments.

Contribution

It introduces a deep active learning approach tailored for avian bioacoustics and discusses key challenges through a pilot study.

Findings

Active learning can significantly reduce annotation efforts.

Deep AL improves model adaptability in diverse environments.

Pilot results indicate promising potential for practical applications.

Abstract

Passive acoustic monitoring (PAM) in avian bioacoustics enables cost-effective and extensive data collection with minimal disruption to natural habitats. Despite advancements in computational avian bioacoustics, deep learning models continue to encounter challenges in adapting to diverse environments in practical PAM scenarios. This is primarily due to the scarcity of annotations, which requires labor-intensive efforts from human experts. Active learning (AL) reduces annotation cost and speed ups adaption to diverse scenarios by querying the most informative instances for labeling. This paper outlines a deep AL approach, introduces key challenges, and conducts a small-scale pilot study.