Contagem automática de ovos de mosquitos Aedes em ovitrampas com visão computacional baseada em aprendizado de máquina

Abstract

The present study proposes the automation of egg counting for Aedes aegypti and Aedes albopictus mosquitoes, vectors of arboviruses, through computer vision based on machine learning. Using images from ovitraps provided by Fiocruz-PE, fine-tuning was applied to advanced models (YOLOv10, YOLOv11, and Faster R-CNN) to overcome challenges related to high variability and noise in the images. The solution reduces manual effort and analysis time, contributing to more effective strategies for controlling diseases such as dengue, zika, and chikungunya.