CNN based Framework for Classification of Mosquitoes based on its Wingbeats

Mosquito is a familiar flying insect which is common in almost all parts of the ecosphere. It has two pairs of thin wings which makes buzzing sound. Wingbeat of a mosquito is a complete set of motions of a wing while a mosquito is in flight mode. The wingbeats differ from mosquito to mosquito. This work mainly concentrates on the wingbeats of mosquitoes and the goal is to build an effective deep learning system which categorizes distinct classes of mosquitoes based on its wingbeats such as Culex (which beats its wings at an average of 400 times in a second), Mansonia (which looks very bigger in size as compared to other mosquito species), Yellow fever mosquito (which causes dengue fever) and Culiseta (which are cold adapted species) using Convolutional Neural Network (CNN). The dataset for this work is obtained from Kaggale voice repository. Residual network-50 (ResNet-50) is used to handle the classification problem of mosquitoes based on its wingbeats and the result is validated based on KFold Validation technique.