First use of unmanned aerial vehicles to monitor Halyomorpha halys and recognize it using artificial intelligence

• Published in: Pest management science Vol. 80; no. 8; pp. 4074 - 4084
• Main Authors: Giannetti, Daniele, Patelli, Niccolò, Palazzetti, Lorenzo, Betti Sorbelli, Francesco, Pinotti, Cristina M., Maistrello, Lara
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01-08-2024; Wiley Subscription Services, Inc
• Subjects: Applications programs; Artificial intelligence; automated pest monitoring; Automation; brown marmorated stink bug; Damage; Drones; Effectiveness; Freezing; freezing behavior; Halyomorpha halys; Image acquisition; Instars; integrated pest management; Mobile computing; Monitoring; Monitoring systems; Pests; Pheromone traps; Unmanned aerial vehicles; automated pest monitoring; drones; brown marmorated stink bug; freezing behavior; integrated pest management
• ISSN: 1526-498X; 1526-4998; 1526-4998
• DOI: 10.1002/ps.8115

BACKGROUND
Halyomorpha halys is one of the most damaging invasive agricultural pests in North America and southern Europe. It is commonly monitored using pheromone traps, which are not very effective because few bugs are caught and some escape and/or remain outside the trap on surrounding plants where they feed, increasing the damage. Other monitoring techniques are based on visual sampling, sweep‐netting and tree‐beating. However, all these methods require several hours of human labor and are difficult to apply to large areas. The aim of this work is to develop an automated monitoring system that integrates image acquisition through the use of drones with H. halys detection through the use of artificial intelligence (AI).
RESULTS
The study results allowed the development of an automated flight protocol using a mobile app to capture high‐resolution images. The drone caused only low levels of disturbance in both adult and intermediate instars, inducing freezing behavior in adults. Each of the AI models used achieved very good performance, with a detection accuracy of up to 97% and recall of up to 87% for the X‐TL model.
CONCLUSION
The first application of this novel monitoring system demonstrated the potential of drones and AI to detect and quantify the presence of H. halys. The ability to capture high‐altitude, high‐resolution images makes this method potentially suitable for use with a range of crops and pests. © 2024 Society of Chemical Industry.
This study demonstrates the potential of unmanned aerial vehicles (UAVs) for automated monitoring, particularly for detecting the mobile stages of invasive species like Halyomorpha halys. The implementation of UAVs and artificial intelligence could optimize insect monitoring by reducing travel time and providing a ready‐to‐use solution to assess the infestation rate of monitored areas.