Action Recognition on American Sign Language using Deep Learning

People who are hard of hearing use sign language to reinforce knowledge they acquire in everyday interactions and to convey their thoughts and feelings. Even though they represent a large portion of any population, the majority of people are unable to communicate with them because they have little to no understanding of sign languages. The goal of sign language recognition is to recognize, evaluate, and comprehend the important hand and body movements. By bridging the deaf community's current communication gap with the rest of society, automatic sign language recognition offers better services to the deaf community. Such devices might save the lives of hearing-impaired individuals who are in life-threatening circumstances like cardiac arrest, crashes, accidents, etc. In this work, deep learning-based models for sign language gesture recognition are created to precisely anticipate the American Sign Language emergency signs (ASL). The collection includes the movies for eight distinct emergency scenarios glosses. The different models used for the study are CNN, LSTM, CNN with LSTM, BiLSTM and LCRN.