DU researchers develop AI-powered device to identify Aedes mosquitoes

Researchers at the University of Dhaka (DU) have developed a device that attracts Aedes mosquitoes using sound, captures their images and uses artificial intelligence (AI) to identify and count them. The researchers believe the technology, named ‘Mast-CloudNet’, could make it easier to monitor the spread of Aedes mosquitoes, which transmit dengue.

The device was jointly developed by researchers from DU’s Department of Electrical and Electronic Engineering and Department of Zoology. A camera inside the device captures images of mosquitoes trapped in it and sends them over the internet to a cloud-based AI system. The AI then analyses the images to identify and count Aedes mosquitoes.

Under conventional methods, mosquitoes must be collected from traps and examined individually in laboratories to identify Aedes mosquitoes. The process is time-consuming and labour-intensive. The new device automates this process.

It attracts male Aedes mosquitoes by producing sounds that mimic the wingbeats of female mosquitoes. Laboratory tests found that sound-equipped traps captured five times as many Aedes mosquitoes as silent traps.

During a preliminary three-day field trial on the DU campus, the sound-equipped device captured 59 mosquitoes, including 49 Aedes mosquitoes and 10 belonging to other species. Aedes mosquitoes accounted for 83.1 per cent of the total catch. By contrast, a silent trap captured only four mosquitoes, including one Aedes mosquito.

The device continued to operate amid rain, wind and the usual noise on the university campus. However, under outdoor conditions, the AI correctly identified 35 of the 49 Aedes mosquitoes, achieving an accuracy rate of about 71 per cent. The researchers said larger-scale trials are needed to further assess the device’s performance. They estimate that it would cost around $99.

The research was led by Mohammad Ahsan Habib, an associate professor in the Department of Electrical and Electronic Engineering; Professor Saeed Mahmud Ullah, from the same department; and Associate Professor Khandaker Ashfaqul Muid of the Department of Zoology. The study was published this year in the journal IEEE Transactions on Instrumentation and Measurement.

Speaking to New Age, Ahsan Habib said the system could provide authorities with timely information about where Aedes mosquitoes are present and how their numbers change over time. This would help officials prioritise the removal of potential breeding sites and assess whether mosquito-control measures are working effectively.

Compared with conventional traps, the device automates mosquito collection and identification, making it a complementary tool for existing mosquito-control programmes, he said.

However, Habib noted that dengue cannot be controlled simply by trapping male mosquitoes. Dengue is transmitted by female Aedes mosquitoes. Although the presence of males can indicate the spread of Aedes mosquitoes in an area, it cannot reliably establish the number of infected females or predict the scale of a dengue outbreak.

The device is primarily intended to monitor mosquito presence and help identify potential breeding sites, he said.

The research team plans to conduct trials using 10 to 15 devices in residential areas under Dhaka South City Corporation in June and July 2027. The trials will focus on improving the AI system’s ability to identify mosquitoes outdoors, protecting the device from weather conditions and assessing its long-term performance. The researchers are also considering the use of solar power and more reliable internet connectivity.

Atike B Zaker, a postgraduate student in the Department of Electrical and Electronic Engineering who led the project and co-authored the research paper, said integrating the hardware with AI software was a major challenge.

The cost of developing the initial prototype increased because of specialised laboratory equipment, component fabrication and imported materials. Zaker said greater access to research funding would help overcome these barriers and facilitate the development of practical technologies.

He added that the team was working on the next version of Mast-CloudNet. Researchers are also testing a potentially cheaper approach that would identify Aedes mosquitoes by detecting the sound of their wingbeats rather than analysing images.