Building Image Database and Retrieval System for Rice Seed Identification

Abstract

       This paper presents a database of rice seeds properties to aid users in the assessment and recognition of seed characteristics of rice varieties based on their apprearance. Inspired by practical demands in quality control and rice seed purity assessment in seed production, we collected rice seeds of different varieties, acquired their images and consulted with experts about the visual properties for rice variety identification. As a result, we built a database of rice seed properties accompanied with their images and the description of the visual features. The database included popular rice varieties in the northern delta of Viet Nam. A database management system (DBMS) was built for managing and searching for rice seeds in the database. A web-based version of the DBMS was also developed. The system can be used to search for different rice varieties based on name and/or the description of the seed appearance. It can help users to learn about rice seed varieties, their visual features for discrimination and recognition. The system is easy to use, especially for non-IT users such as technicians, agronomy students or rice experts of rice seed production companies. This is also the first web-based database of rice varieties in Viet Nam with rice seed images to support rice seed searching and recognition.