Jannelle Couret of the University of Rhode Island, USA, trained a Convoluted Neural Network (CNN) using 1,709 two-dimensional adult mosquitos images. The mosquitoes were collected from different colonies in five geographic regions and stored by flash freezing or dried samples. The researchers trained the CNN to identify Anopheles mosquito and its sex. Using test data, researchers found a 99.96% prediction accuracy for class and a 98.48% accuracy for the sex of the mosquito.
Identification of mosquitoes is complicated and nearly indistinguishable even to trained taxonomists. However, AI-based rapid identification of mosquito can potentially improve mosquito surveillance, which is essential in the fight against human pathogens such as malaria.