Classification of Thai Rice – Grain Images using Mask R – CNN and Transfer Learning

Classification of Thai Rice – Grain Images using Mask R – CNN and Transfer Learning

Faculty of Information and Communication Technology, Mahidol University

Project Title:

Classification of Thai Rice-Grain using Image Processing

Research Title:

Classification of Thai Rice – Grain Images using Mask R – CNN and Transfer Learning

Researcher (s):

Asst.Prof. Dr.Worapan Kusakunniran
Mr.Kittinun Aukkapinyo
Mr.Parintorn Pooyoi
Mr.Suchakree Sawangwong

Rice is the main part of each meal in Thailand. There exist differences in characteristic and appearance on each type of rice grains. At seed trading at Thai rice mill, there exists a process to verify the correctness of rice grains in each bag of rice. Currently, the process is sampling some of the rice grain in each bag randomly and a human expert in each mill will verify the correctness of sampled rice grains. However, there are no human experts to classify rice grains in every mill and their number tends to decrease as time passed by.

From the problems mentioned above, we aim to propose a framework to develop a data model that can classify and localize each rice grain in an input image. The framework is mostly an iterative process which consists of data acquisition, data preparation, data modelling, and model evaluation. Then, an application with a simple user interface will be developed in order to visualize the classification result.

Award Grant related to the Project:
• Silver Medal at Thailand Research Expo 2019 by National Research Council of Thailand

Key Contact Person:
Asst.Prof. Dr.Worapan Kusakunniran
Faculty of Information and Communication Technology, Mahidol University