Contactless Sensing for Sign Language Phrase Recognition

Sign language (SL) is an essential means of communication for people who are deaf or have hearing impairments. Despite its significance, a language barrier persists between the deaf community and individuals unfamiliar with SL. This paper explores the potential of Sign Language Recognition (SLR) technology to bridge this gap by focusing on phrase-level recognition using deep learning (DL) models. A dataset consisting of 8 common phrases—Again Please, Bless You, Excuse Me, Good Afternoon, Good Morning, Hello, Welcome, Slow Please, and Stay Safe was collected using the Xethru X4M03 Ultra-Wideband (UWB) radar sensor, with 240 samples across 8 classes. Two pre-trained DL models, including VGG16, and GoogleNet were applied to the dataset to classify the phrases. The performance of these models was evaluated based on accuracy, precision, recall, and F1 Score. Results showed that VGG16 outperformed GoogleNet, achieving an accuracy of 80%. This study demonstrates the feasibility of using radar-based contactless sensing and DL models to improve communication for the deaf and hard-ofhearing community.