Mango Leaf Disease Identification using a Fuzzy Neural Approach

Mangoes are one of the main crops produced in the Philippines. Once infected, the fruit produced by the tree can become less likely to be sold or be unsell able. The study aims to detect three diseases in the mango tree leaf: anthracnose, dieback, and sooty mold, to help identify diseases that can affect harvest. The study makes use of both fuzzy and convolutional neural networks to identify mango diseases. Images of mango leaf samples were gathered from MangoLeafDB with five hundred (500) samples from each category were used in training the system with the Python Programming Language. One hundred twenty-five (125) leaves were used in testing the model with twenty-five (25) samples from each category of anthracnose, die back, sooty mold, healthy, and unknown leaves. Upon using a confusion matrix, the system's accuracy is determined to be 92%.