Building Ensemble of Resnet for Dolphin Whistle Detection

Ecoacoustics is arguably the best method for monitoring marine environments, but analyzing and interpreting acoustic data has traditionally demanded substantial human supervision resources. These bottlenecks can be addressed by harnessing contemporary methods automated audio signal analysis. This paper focuses on problem of assessing dolphin whistles using state-of-the-art deep learning methods. Our system utilizes a fusion various resnet50 networks integrated with augmentation (DA) techniques applied not to training test set. We also present speeds classification results DA Through extensive experiments conducted publicly available benchmark, our findings demonstrate that ensemble yields significant performance enhancements across several commonly used metrics. For example, approach obtained an accuracy 0.949 compared 0.923, reported in literature. provide testing sets other researchers use comparison purposes, as well all MATLAB/PyTorch source code this study.