Mobile Sign Language Interpretation Using Inception Ver. 3 Classifier

Abstract

Communication plays a key role on a daily basis. An essential skill to possess is the capacity to articulate thoughts and ideas in a straightforward and succinct manner. People who have a basic understanding of the language can speak it fluently. However, individuals who struggle to articulate their thoughts verbally have a contrasting situation. Individuals who lack the ability to vocalize communicate through sign language, which is not universally comprehensible. Occasionally, this can be a hindrance to their everyday activities. For instance, when individuals with hearing impairments engage in communication via gestures, it captures the interest of those involved in the field of human-computer interaction, who then closely scrutinize every aspect of this remarkable setting. In contemporary times, it appears that nearly everyone possesses a smartphone and has the ability to access the internet. With the continuous advancement of technology and connectivity, an increasing amount of data is becoming readily available. This can be employed in conjunction with deep learning to train a model to comprehend the indicators. Our model’s proficiency in interpreting various sign languages will significantly enhance communication between the general population and individuals with hearing impairments. We have developed a smartphone application that utilizes deep learning algorithms to recognize and classify the gestures of an individual, subsequently presenting the relevant textual information. The proposed model makes use of the well-known and extremely efficient Inception v3 architecture to perform picture identification and classification tasks. Inception v3 performed admirably in the ImageNet Large Scale Visual Recognition Challenge (ILSVRC), proving itself as a cutting-edge model.