Sign Language Recognition Using LSTM Model: A Comparative Analysis of CSL and ArSL Datasets

Enhancing communication between the deaf community and the people outside makes Sign Language Recognition (SLR) crucial element in present day research. We present a deep learning model for Sign Language (SL) alphabet recognition utilizing Long Short-Term Memory (LSTM) networks. This SLR system is trained and tested for two open source sign language alphabets datasets: Arab Sign Language (ArSL) and Chinese Sign Language (CSL). Our analysis shows that the model achieves an outstanding accuracy of 89.63% and 92.03%, on ArSL and CSL datasets, respectively. Higher values of the accuracy projects the usefullness of our LSTM model for possible real-time applications pertaining to development of improved communication as well as assistive tools for the deaf community.