A Deep Neural System for Mosquito Classification

Abstract

In recent years, the realms of Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) have experienced a transformative evolution, reshaping numerous domains across industries and academia. This revolution has been driven by the exponential growth in computing power, the availability of vast amounts of diverse data, and advancements in algorithms and methodologies. Entomology is one such domain that emerges as a field significantly influenced by the effective application of ML and DL techniques. These techniques offer innovative strategies for Vector Surveillance and Control, Automated Species Identification, and Insect Population Monitoring. newlineInsect-borne diseases, with mosquitoes being a prominent vector, continue to pose significant public health challenges globally, necessitating efficient and proactive control measures. Mosquitoes serve as primary vectors for life-threatening diseases, including malaria, dengue, West Nile virus, chikungunya, yellow fever and Zika fever. Mosquitoes primarily belonging to three genera - Aedes, Anopheles and Culex play crucial roles in the transmission of diseases to humans. According to WHO (World Health Organization) reports, Mosquito-borne diseases (MBDs) account for more than one million cases annually worldwide. Given the lack of safe and effective vaccines or treatment for MBDs, vector population control remains the major approach for controlling these diseases. newlineSurveillance of vectors is critical to monitor and identify the mosquito species prevalent in a geographic area in real-time to implement effective vector control interventions. Existing vector surveillance approaches are labor-intensive, often resulting in unreliable surveillance data due to delays, misclassification, and spatial constraints, particularly in remote areas. These challenges underline the need for developing automated mosquito classification techniques.