Rice (Oryza Sativa) Grading classification using Hybrid Model Deep Convolutional Neural Networks - Support Vector Machine Classifier

Abstract

Rice grading plays an essential role in identifying the rice production industry's rice quality method, including its market price. Rice quality is one of the critical selection criteria highly prioritized by farmers and rice consumers, primarily determined by its different rice characteristics. This research paper focuses on developing a hybrid model in classifying rice milled grading: Premium, Grade 1–5 using Raspberry Pi microcomputer based on the physical features extracted such as damaged, discolored, broken, and chalky rice grains present in the sample by integrating the key properties of Deep Convolutional Neural Networks (DCNN) for feature extraction and Support Vector Machine (SVM) as a classifier. An enclosed staging platform with constant and uniform illumination was used for image acquisition with 150 grains per image sample. The proposed model has identified and classified rice grading proficiently and achieved a classification training and validation of 98.33% and 98.75%, respectively.