3D-CNN Empowered Assistive Machine Learning Model for the Hearing Impaired

According to the World Report by WHO on Hearing published in March, 2021, nearly 2.5 billion people will be deaf or hard of hearing by the year 2050. In 2020, the uncorrected hearing loss in the world costs about $1 trillion [1]. In order to communicate effectively with persons who are deaf or hard of hearing, sign language serves as an interface. Hand, finger, arm, head and facial expressions are all part of a sign language that includes gestures and symbols. Using visual descriptions, they are able to better understand the world around them and hence contribute to society. In today's world of impulsive technologies, building an assistive sign learning paradigm is a significant issue, particularly because many sign language apps are ineffective at identifying the signs they are intended to teach [2]. In the proposed framework, gestural recognition is being performed using a 3-D Convolutional Neural Network (3-D CNN) [3], [4]. Once trained, it can recognize patterns in volumetric data like videos and reliably identify the features it needs. Although, it is intended for dumb and deaf persons to use sign language to communicate with others, the proposed system seeks to comprehend some key features of sign language and convert them into text or voice using 3-D CNN. Intermediary translators are no longer needed as a result of this change. To this aim, this paper proposes a model which will enable hearing impaired persons to communicate easily with a world around them.