Construction of SIBI Datasets for Sign Language Recognition Using a Webcam

Abstract

SIBI (Sistem Isyarat Bahasa Indonesia) is a sign language established by the Government of Indonesia as a standard of communication that used in Special School. Hence, not everyone can understand the sign language, even though people have ability to hear. They also must be able to understand the sign language in order to conduct a communication. Currently, technology of recognition has been applied to hand gestures that can translate sign language. Various studies regarding sign language recognition have been developed and there are many applications that can recognize sign language through hand gesture as datasets using webcam, but there still restricted on the datasets in SIBI. Therefore, the study aims to create SIBI datasets via hand gesture using experimental method. So that, the application can be used by SIBI users. SIBI datasets were successfully created with signs that consisting of 8 static word. Based on the confusion matrix, all the signs that has been created have a precision of 100%, meanwhile based on the test results with sufficient lighting conditions and plain background, SIBI datasets has a prediction success rate of 97,5%.