Pengenalan Alfabet SIBI Menggunakan Convolutional Neural Network sebagai Media Pembelajaran Bagi Masyarakat Umum

Abstract

SIBI is one of the Sign Languages used in Indonesia and has been widely used in the community, especially the school (SLB). Communication limitations of the deaf and speech community cause limited communication with the general public, especially many general public who do not know Sign Language or SIBI. For this reason, this research was conducted in order to become a learning media for the general public in recognizing the SIBI alphabet so that it can support communication with the deaf and speech community. This research was conducted to become a medium that can be used as a learning medium in the introduction of the SIBI alphabet. The method used in this research is CNN. CNN is used because it is a deep learning method that has the most significant results in image recognition. The data used is 2,600 images which are divided into 80% training data and 20% validation data. Training was done ten times by comparing the parameters that produce the best accuracy. The parameters used are batch size and epoch. From ten trials, the best accuracy is obtained using batch size 8 and epoch 50. The best accuracy produced is 85% training accuracy and 87% validation accuracy.

Comparative Analysis of CNN Architectures for SIBI Image Classification

“…In their research, Fadhilah and Marpaung employed the Convolutional Neural Network (CNN) method, known for its effectiveness in deep learning for image recognition, to develop a learning medium for recognizing the SIBI alphabet [7]. The dataset used consisted of 2,600 images, divided into 20% for validation and 80% for training.…”