BISINDO (Bahasa Isyarat Indonesia) Sign Language Recognition Using Deep Learning

Abstract

<p>Sign language recognition plays a crucial role in<br />facilitating communication for individuals with hearing<br />impairments. This paper presents a deep learning-based<br />approach for recognizing Bahasa Isyarat Indonesia (BISINDO),<br />the sign language used in Indonesia. The proposed system<br />employs convolutional neural networks (CNNs) and recurrent<br />neural networks (RNNs) to automatically extract features from<br />sign language gestures and classify them into corresponding<br />linguistic units. The dataset used for training and evaluation<br />consists of annotated BISINDO sign language videos.<br />Preprocessing techniques such as normalization and<br />augmentation are applied to enhance the robustness of the<br />model. Experimental results demonstrate the effectiveness of the<br />proposed approach in accurately recognizing BISINDO sign<br />language gestures, achieving state-of-the-art performance<br />compared to existing methods. The developed system shows<br />promising potential for real-world applications in enhancing<br />communication accessibility for the hearing-impaired<br />community in Indonesia.</p>