Image-based Approach to Detect Bakanae Disease on Rice Seedlings

Abstract

Abstract. Bakanae, also known as foolish seedling, is a threatening disease to rice (Oryza sativa L.). Infected plants can yield empty panicles, causing reduction in grain production. The disease can infect rice grains at storage, and can spread in the field. It is essential to screen infected plants at their early stage. Conventional methods for screening the infected plants are laborious and destructive. This work proposed an image-based approach to differentiate infected and healthy seedlings at the age of 3 weeks. In the experiment, grains of a rice cultivar Tainan 11 were inoculated with the pathogen and then cultivated in an incubator for 3 weeks. The infected seedlings were photographed. Morphological and color traits of the seedlings were quantified using image processing algorithms. Support vector machine classifiers were developed to distinguish the infected and healthy seedlings. It was demonstrated that the proposed approach could identify diseased seedlings at an accuracy of 88.33%.