Kangwon National University professor develops 1st AI detector for bee mites

A photo of BeeSion, an artificial intelligence (AI)-powered varroa mites detector / Courtesy of Rural Development Administration

A research team led by Mo Chang-yeon, professor at Kangwon National University's Bio System Engineering Department, developed the world's first artificial intelligence (AI)-powered device to detect varroa mites, in collaboration with the Rural Development Administration (RDA).

Professor Mo Chang-yeon at Kangwon National University (KNU) / Captured from KNU website

Varroa mites are parasitic pests that attack and feed on honeybees inside the hive, causing developmental damage and transmitting deadly viruses. They are considered the most destructive pest threatening honeybees worldwide and are recognized as the leading cause of mass honeybee die-offs during winters.

The device, BeeSion, analyzes videos of hive frames using AI to find varroa mites in real time. Varroa mite control is extremely difficult, since the pests reside inside hives and are hard to detect, especially during summer.

While traditional methods require over 30 minutes to inspect a single hive, BeeSion only takes about 30 seconds per frame and has an accuracy rate of 97.8 percent.

The device can also simultaneously detect 16 other types of bee diseases and abnormalities, including chalkbrood infection, wing deformities and larval disorders, and has notification guidelines depending on the level of infestation, from "recommended for monitoring" to "caution" to "intensive treatment."

“This research achievement will play an important role in reducing the labor burden for beekeepers, improving pest control efficiency and safeguarding the stability of domestic agriculture and food security. We will continue to work with the Rural Development Administration to further advance AI-based smart beekeeping technology and present a world-class model for digital transformation in agriculture.”

During the device’s testing, about 150 farms using BeeSion saved about 8.6 million won ($6,200) annually, recouping the device’s cost of 4 million won within a year.

By detecting pests and disease symptoms in earlier phases, the device is expected to help reduce colony losses significantly, as virus transmission and resistance to chemical treatment for varroa mite infestation are key factors in honeybee loss.

Mo and the RDA plan to share the devices’ technology with the industry this year to commercialize production and roll it out to beekeepers starting in 2028.

“This achievement marks the first case of applying data and AI technology to beekeeping, which has traditionally relied on experience. It will serve as a turning point toward precise management of hives and automated pest monitoring,” the RDA said.

“Moving forward, we will further develop proactive and digital-based monitoring systems to protect honeybees and strengthen the competitiveness of the beekeeping industry.”