Adaptive Neuro-Fuzzy Approach for Cacao Bean Grading Classification Process

In the Philippines, the current strategy of fermented cacao bean classification is somewhat particular and restricted in human's capability to calculate bean quality. Thus, the study digitized the process using image processing and the Adaptive Neuro-fuzzy Inference system technique (ANFIS). There are two phases involves in the process; the first phase is the input bean classification- to identify and classify the bean input sample if it is moldy, slaty, and defected using the image processing technique. The second phase is the Grading process to determine the samples submitted for categorization according to their grade level value (Grade 1A, 1B, 1C, 2A, 2B, 2C, and Sub-standard) using the Adaptive Neuro-Fuzzy Inference System. There were two hundred data sets fed to MATLAB software using the ANFIS technique for the Grading classification process. This study utilizes eighty (80%) percent of the samples for training, and for checking, it employs the remaining twenty percent (20%). During the training, the accuracy rate projected value resulted in 99.71%, and the said technique outperforms the KNN method for cacao bean sorting and grading classification. With the results obtained, the ANFIS technique is a technique that can be applied effectively to categorize the quality standard of the cacao bean samples submitted for the Grading categorization process.