The Japanese maple scale (JMS) (Lopholeucaspis japonica) is an armored scale insect that causes significant losses in the ornamental industry through direct injury, plant rejection, unmarketable stock, increased pest control costs, and restricted interstate plant movement. Ornamental growers typically identify JMS by visually inspecting trunks, branches, and twigs for waxy covers or by using sticky tape traps to monitor crawler activity. However, its microscopic size and camouflaged appearance make early detection challenging. This study aims to develop an artificial intelligence (AI)-guided, on-the-go pest scouting system to address the current challenges in early management of JMS in ornamental crop production. To develop the scouting system, a Sony ILX-LR1 professional camera with a 61.0 megapixel full-frame sensor and interchangeable E-mount lenses designed for detailed industrial applications is used to capture high-resolution images. A small amount of data has been collected so far, with plans to gather a larger image dataset during the summer months. In the initial analysis, captured images were sliced into smaller patches to make the microscopic JMS detectable. These sliced images were used to train a transformer-based AI model for detecting JMS. The trained model, tested on the small dataset, showed it could detect JMS with an Intersection over Union (IoU) of over 0.8. While the model shows potential for detecting microscopic JMS, comprehensive training and testing with a larger dataset are needed to validate its performance. Upon completion, the developed scouting technology will serve as an effective tool for early detection and management of JMS in nursery environments, reducing plant injury and rejection while improving profits for ornamental growers.
