Classification of rice grains obtained from processed rice paddy with machine learning

Nowadays, the classification of the rice grains separated from the husk by pulling the rice paddy is made by hand. This method results in economic loss by causing time loss and labor inefficiencies in the related sector. In order to produce a solution to this problem, in this study, it is aimed to classify previously extracted rice grains by machine learning method. Machine learning method has been used for this classification. With this method, machines are able to perform specialized tasks such as classification, object recognition, voice recognition or language translations with the least error by using data such as voice, image, signal and text. In recent years, with the continuous development of machine learning methods, this method has found an important place in fields such as image processing, sound processing and robotics. With this thesis, it is aimed to eliminate time loss and labor inefficiency in the related sector by using machine learning method and to provide the basis for the studies that will contribute to the elimination of existing problems in other sectors that need to be classified in the following processes. In this thesis, the process of photographing and shredding of rice is explained in detail. Sobel Edge Detection Algorithm was used in the process of fragmentation. In order to obtain the properties of the images of rice, detailed information about size estimation, determination of color spaces and values of RGB, HSL, HSV, CIEXYZ, CIELAB and GLCM feature vector were explained. Machine learning designs, which give its name to this thesis, are Decision Trees, Support Vector Machines, Naive Bayes, K-Nearest Neighborhood Algorithm and Deep Learning Method. The information about these models, how they are used in practice and the results related to the application are shared in the last section. At the end of the thesis, the results of machine learning method are shown comparatively and the most successful network model is Quadratic Support Vector Machines.