Low Resource Species Agnostic Bird Activity Detection

TL;DR

This paper presents a low-resource, species-agnostic bird activity detection system optimized for embedded devices, using lightweight features and classifiers to achieve competitive accuracy with reduced computational cost.

Contribution

It introduces a novel combination of low-level spectral, pitch, and amplitude modulation features with lightweight classifiers for bird activity detection on resource-constrained devices.

Findings

Random forest achieved 0.721 accuracy and 0.604 F1-Score.

Features and models with lower computational cost perform comparably or better than CNN-based detectors.

The system is suitable for edge deployment in long-term bird monitoring.

Abstract

This paper explores low resource classifiers and features for the detection of bird activity, suitable for embedded Automatic Recording Units which are typically deployed for long term remote monitoring of bird populations. Features include low-level spectral parameters, statistical moments on pitch samples, and features derived from amplitude modulation. Performance is evaluated on several lightweight classifiers using the NIPS4Bplus dataset. Our experiments show that random forest classifiers perform best on this task, achieving an accuracy of 0.721 and an F1-Score of 0.604. We compare the results of our system against both a Convolutional Neural Network based detector, and standard MFCC features. Our experiments show that we can achieve equal or better performance in most metrics using features and models with a smaller computational cost and which are suitable for edge deployment.