Scientists have developed a method to investigate and control disease-transmitting tiger mosquitoes This takes the form of a deep neural network architecture that is capable of identifying the tiger mosquito from images.
This feat was performed within the Scene understanding and artificial intelligence (SUNAI) research group, based at Universitat Oberta de Catalunya. The machine learning process was achieved by processing and analysing a large number of images that volunteers took using mobile phones. The images were uploaded to the Mosquito Alert platform.
The Mosquito Alert system is a non-profit cooperative citizen science project, coordinated by different public research centers. The objectives are to study, monitor, and fight the spread of invasive mosquitoes capable of transmitting global diseases. Such diseases include dengue, Zika, or West Nile fever.
The tiger mosquito (Aedes albopictus) comes from the mosquito (Culicidae) family. Thus mosquito is found in the tropical and subtropical areas of Southeast Asia. The insect is characterized by the white bands on its legs and body. The mosquito is a disease vector for yellow fever virus, dengue fever, Chikungunya fever, and Usutu virus.
This mosquito also closely associates with humans and it typically flies and feeds in the daytime. It is of concern that rising temperatures worldwide are facilitating the spread of the insect and associated diseases.
In developing the project, researchers encouraged volunteer citizens to send in images. The data was coupled with data relating to breeding sites in public spaces. The information was processed by entomologists.
The reason why the artificial intelligence is important is because identifying mosquitoes helps to address disease transmission. It remains a challenge for researchers to correctly identify the type of mosquito.
One means is within the laboratory by analysing the spectral wave forms of mosquito wing beats, the DNA of larvae and morphological parts of the body. This is very time-consuming. The method presents a more streamlined approach.
This is because a feature of neural networks is that they can be trained through supervised, semi-supervised, or unsupervised manner to process data and guide the network about the type of result being sought. In addition, large volumes of data can be processed.
In trials, the neural network developed can perform as well or nearly as well as a human expert and the algorithm is sufficiently powerful to process massive amounts of images.
The research appears in IEEE Access and it is titled “A Deep Convolutional Neural Network for Classification of Aedes Albopictus Mosquitoes”.
