Rice Grain Classification Using Machine Vision with Raspberry Pi

Lee, Ze Lim (2022) Rice Grain Classification Using Machine Vision with Raspberry Pi. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Lee Ze Lim.pdf Restricted to Registered users only Download (8MB)

Abstract

Rice is a major food all over the world and fulfills the needs of a dominant portion of people. Rice quality is determined by various parameters including length, colour, and others. In recent years, there has been an increasing demand for smart rice cookers. An intelligent and efficient system that can automatically identify the cooking time and amount of water by recognizing the rice grain inside the rice cooker is required. In this research, a system that can automatically recognize and classify rice grains based on their respective types by using machine vision technology with a raspberry pi processor is proposed. The collection of datasets is the main process in the development of the system. A total of 700 pictures of rice grains, 100 images of each different type of rice grain are included in the dataset. This project involves the extraction of various physical parameters of individual rice grains such as length and colour. This study involves 7 types of rice grains they are white long grain, white medium grain, white short grain, 10-grains rice, black rice, red rice and brown rice. Artificial Neural Networks (ANN) and Convolutional Neural Networks (CNN) are used to develop classification models. The performance of ANN and CNN models is evaluated by using performance metrics. From the average classification accuracy, the highest classification accuracy belongs to the CNN method, which is 99.52%. The ANN model has achieved an 88.16% accuracy which is slightly lower than the CNN model. The real-time simulation process of the CNN model and ANN model is 3s and 0.5s respectively. A proof-of-concept for the smart rice cooker is developed using a raspberry pi microcontroller.