Assessing narrow brown leaf spot severity and fungicide efficacy in rice using low altitude UAV imaging

Abstract

The utilization of unmanned aerial vehicle (UAV)-based imaging systems offers precise detection of plant diseases and aids in decision-making regarding fungicide applications to optimize disease control. This study employed a twin-lens multispectral camera mounted on a low-altitude UAV to capture imagery data from rice field plots that were treated with different fungicides. The objectives of this study were to assess the severity of narrow brown leaf spot (NBLS) caused by Cercospora janseana and to evaluate the efficacy of fungicide control. Eighteen color features and vegetation indices were extracted from RGB images, while nine color features and vegetation indices were extracted from multispectral images. Through correlation analysis, four spectral features, namely Lab-a, ExGR, VDVI, and g, were found to exhibit high correlations with disease severity. Specifically, RGB imagery had greater correlation coefficients (exceeding 0.95) for both ExGR and Lab-a features compared to multispectral imagery. A multifeature inversion modeling approach was employed, using support vector regression with the top four spectral features to predict NBLS severity. The results indicated R2 values were above 0.93 for all support vector regressions. Furthermore, the efficacies of ten different fungicide treatments were evaluated, with UAV imaging consistently aligning with ground truth rating data in terms of efficacy ranking. These results demonstrate the potential of UAV imagery for use as a valuable tool for NBLS detection and assessing fungicide efficacy, offering significant benefits in the management of NBLS, which is a globally important disease in rice.