Development of Mobile Application for Automatic Identification of Biotic Diseases in Oryza Sativa Using Image Processing Techniques

摘要

Oryza sativa, commonly known as Asian rice, is the most important staple crop in the Philippines. Farmers are experiencing yield losses on their rice crop due to damages caused by certain types of biotic diseases. It is necessary to determine the specific disease infesting the crop in order to apply the proper cure at an early stage. Existing methods for biotic disease identification include laboratory molecular analysis and a diagnostic tool mobile application such as the Rice Doctor. Laboratory methods have relatively higher cost and higher turnaround time, while rice doctor is complex to use according to farmers' perspective. This paper presents the development of Rice Checkup, an offline mobile application capable of automatically identifying common biotic diseases found in oryza sativa namely bacterial leaf blight, tungro, sheath blight, and rice blast. The aforementioned app uses algorithms devised from Otsu's method, RGB, and color and shape recognition. A total of 385 samples, equally distributed among types of biotic diseases, were gathered from the experimental sites located inside the International Rice Research Institute (IRRI) and University of the Philippines Los Baños (UPLB). These samples were infected by known pathogens causing the respective biotic diseases covered in this study. An average identification accuracy rate of 90.91% at a reliability of at least 95% was obtained using a total of 1,046 test images. The solution presented here is practical, inexpensive, fast, easy to use, and available in local (Filipino) language.