Utilizing Support Vector Machines to Recognize Pattern Numbers on Sheet C1 Ballots

Abstract

The election vote-counting process can be done automatically by applying machine learning technology, namely computer vision which is usually used for the pattern recognition process in images and videos. In this study, the object of pattern recognition is the C1 ballots for the 2019 simultaneous elections. The dataset used is the latest C1 format which was made by the author himself. To distinguish each number, C1 ballots will be colored to interpret a number. Then the numbers were captured using an android camera and saved in jpeg format. Furthermore, the author uses the python skimage module and GIMP for image processing and divides the dataset into 80% as training data and 20% as testing data. The model used in this study is the Support Vector Machine algorithm to classify images. The results that can be reached by this model were 100% accurate.