Thermal image evaluation of crop diseases in the incubation period.

Abstract

Early blight is one of the main diseases that causes serious pepper yield and quality reduction. The identification of the presymptomatic features during the incubation period is of great research significance, in that scientific prevention measures in the incubation period may effectively reduce the loss of peppers. The present study confirms the feasibility of the identification of presymptomatic features in pepper early blight by infrared thermography during the incubation period. The infrared thermal images of pepper blades were captured during the whole incubation period in our experiment. Evaluation parameters such as temperature uniformity (U) and the variation coefficient (C) were established by infrared thermal images for the blade temperature field. Evaluation parameters such as temperature uniformity U and variation coefficient C of pepper blades change during the time from being inoculated to significant morbidity. For cases of stabbing inoculation, there were no relatively obvious symptoms in visible light images until 96 h after inoculation, whereas some presymptomatic features appeared in infrared thermal images at 60 h after inoculation. Based on the uniformity changes (δU) and the variation coefficient changes (δC) in the temperature field distribution, the characteristic polynomial (CP)-GBDT model has been established by optimizing the gradient boosting decision tree (GBDT) model. Compared with the GBDT mode, the CP-GBDT model accuracy increased from 87.5% to 91.7% on the test set. Evaluation parameters such as temperature uniformity U and variation coefficient C can be used to effectively characterize the presymptomatic features of pepper early blight by infrared thermography during the incubation period. The results of the present study can provide a reference for the detection of crop diseases in the incubation period. © 2024 Society of Chemical Industry.