Signal classification by similarity and feature extraction with application in automatic insect identification

Abstract

Insects have a strong relationship with the human-beings. For example, some species of mosquito transmit diseases that kill millions of people around the world. At the same time, the presence of certain insects is essential for the ecological balance and food production. For this reason, we are developing a novel sensor as a tool to efficiently control disease vectors and agricultural pests without harming other species. In this paper, we demonstrate how we overtook the most important challenge to make this sensor practical: the creation of accurate classification systems. Despite the short duration and the very simple structure of the signal, we managed to successfully identify relevant features using speech and audio analysis techniques. We show that we can achieve an accuracy of 98% in the task of disease vector mosquitoes identification.