Analysis of Human Visual Behavior between Experts and Novices during Diagnosis of Citrus Greening Disease (HLB)

Citrus greening disease, also known as Huanglongbing (HLB), causes the leaves of citrus trees to gradually yellow and eventually die. HLB is caused by a plant pathogenic bacterium that spreads during the grafting of infected trees. The primary control method of HLB is to cut down the infected trees. Therefore, early diagnosis is essential. Conventionally, the primary method for identifying infected trees was the polymerase chain reaction (PCR). However, PCR tests are time-consuming and expensive. Deep-learning-based HLB diagnostic systems have been proposed; however, their diagnostic accuracy for the entire tree remains low. Therefore, this study aims to quantitatively clarify the differences in gaze movements between experienced and novice observers during disease diagnosis, thereby contributing to the identification of affected areas. Moreover, we conduct eye-tracking experiments to examine differences in visual behavior between experts and novices. The results suggest that experts are more likely to focus on and examine regions relevant to disease identification than novices, indicating that visual behavior reflects expertise in visual disease diagnosis.