Comparative Study Between CNN and LSTM Approaches for Sign Language Recognition

Image recognition technology has advanced to the point where it can help address challenges such as communication barriers faced by individuals with hearing impairments. Due to the societal rarity of learning and practicing Sign Languages (SLs), Machine Learning (ML) techniques have come to light to ease and automate this process through Sign Language Recognition (SLR) systems. This paper compares two SLR approaches; namely: Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM). These two ML approaches are used to accurately identify the English alphabet, numbers, and specific words in American Sign Language (ASL). Results show that both techniques achieve very good accuracy values using test dataset, however in real-time tests CNN has outperformed LSTM.