Estimation of cercospora leaf spot disease severity in table beets from UAS multispectral images

Cercospora leaf spot (CLS), caused by the fungus Cercospora beticola Sacc., is a severe foliar disease that affects the health of table beet crops. Remote sensing techniques, specifically unmanned aerial systems (UAS), could offer a valuable solution for evaluating disease progression and targeted fungicide application towards effectively managing CLS. We conducted UAS flights over C. beticola inoculated table beet plots at Cornell Agritech in Geneva, NY, USA during the growing seasons of 2021, 2022 and 2023. Five-band (475,560, 668, 717 and 840 nm) multispectral imagery was collected for the 2021 and 2022 flights, while for 2023 flights, a four-band (560, 650, 730 and 860 nm) sensor was used. CLS disease severity was assessed in situ at regular intervals, severity was evaluated using both spectral and textural features. Vegetation indices, derived from multispectral imagery, are known to be effective in assessing plant health and disease determination. Texture features also have been reported to be effective in determining plant traits since they provide additional pixel-spatial information to the model. Extensive selection was used to obtain the most relevant features. Our best-performing machine learning model used a random forest regressor to estimate disease severity at an R2test = 0.82 and RMSEtest = 9.31%. Furthermore, the disease progression was tracked using area-under-disease- progression-stairs (AUDPS) and the predicted values had a correlation coefficient of 0.84, 0.94 and 0.81 with 2021, 2022 and 2023 data, respectively. These results are strongly encouraging to support table beet growers with advanced data collection and analysis methodologies, enabling them to implement rapid and efficient foliar disease management strategies.