Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Deep Active Learning in Avian Bioacoustics

Lukas Rauch, Denis Huseljic, Moritz Wirth, Jens Decke, Bernhard Sick, Christoph Scholz

TL;DR

This paper outlines a deep AL approach, introduces key challenges, and conducts a small-scale pilot study on passive acoustic monitoring in avian bioacoustics.

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.

Towards Deep Active Learning in Avian Bioacoustics

TL;DR

This paper outlines a deep AL approach, introduces key challenges, and conducts a small-scale pilot study on passive acoustic monitoring in avian bioacoustics.

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.
Paper Structure (6 sections, 2 figures)

This paper contains 6 sections, 2 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Active Learning in Bird Sound Classification
  4. Experiments
  5. Open Challenges and Limitations
  6. Conclusion

Figures (2)

  • Figure 1: Proposed deep AL cycle in avian bioacoustics with exemplary tasks from BirdSetrauch2024birdset.
  • Figure 2: Improvement curves of deep al selection strategies Badge, Entropy, and Typiclust over Random with the metric collection a) AUROC, b) cmAP and c) T1-Acc. The results are averaged over ten randomly initialized repetitions to ensure consistency and the standard deviation is displayed.