Modified Horsfall-Barratt-based Mamdani Fuzzy Logic System for Philippine Maize Leaf and Ear Disease Diagnosis

Plant diseases are a risk to the efficiency of agriculture. Pathogen-infected plants produce less yields and become malnourished, lowering economic viability. To treat these diseases, however, requires accurate identification of responsible pathogens to perform proper treatment procedures. In this study, Mamdani fuzzy logic system was developed as a fast method to complement slower but precise methods that require chemical analysis. The system distinguished between three ear (Aspergillus, Fusarium, and Gibberella) and five leaf diseases (Southern Corn Rust, Maize Dwarf Mosaic, Southern Corn Leaf Blight, Philippine Downy Mildew, and Banded Leaf and Sheath Blight) found in maize (Zea mays), an important cereal crop harvested for its grains, based on leaf marking discolorations and fungal ear growth colors through modified Horsfall-Barratt (HB). The disease severity was diagnosed based on leaf marking elongation and infected area percentage. MATLAB was used for the implementation and testing of the fuzzy systems against a synthetic dataset. Testing yielded accuracies of 90% for maize ear disease identification, 75% for maize ear disease severity, 82.5% for maize leaf disease identification, and 69.9% for maize leaf disease severity. The system has potential in being implemented as a support tool for farmers to identify diseases in plantations without the need of chemical-based methods for faster treatment response times.