A Machine Learning based Insect Bite Classification

Abstract

Identifying insects by their bite marks can assist doctors in diagnosing victims and providing appropriate treatment. In recent years, researches using Machine Learning have been actively conducted and have produced excellent results in fields such as object detection, behaviour recognition, voice recognition, and cancer detection in medical field. This study has developed a classification application that can be used on mobile phones to solve the insect classification problems. Experiments were carried out on five insect species chosen for being the most common biting insects. Detailed study was conducted on different images with the help of Random Forest and Support Vector Machine models. These models need different insect bite marks images to classify them. Random forests achieve a better performance and are usually much faster than Support Vector Machines.