Flex Sensor Dataset: Towards Enhancing the Performance of Sign Language detection System

Abstract

Whether it's God's creation or science, the earth has evolved and life came into existence. Not all have the same character instead exhibit similar. As of reports from the National Institute on Deafness and Communication related disorders 18.5 million(approx.) have speech related disorders. Many are born with speech inability and many get speech-related disorders and are unable to speak. As communication is key for an effective society and environment, a person with speech inability looks different from a normal person. A communication gap exists between a normal person and a person with speech inability. As technological revolutions have taken the world to a new level, still a better and efficient process to make a speech disorder person a normal person is not available. There are many processes and products available but the efficient and reliable ones are not available. The use of Machine Learning and Artificial Intelligence in many fields has high success rates. In this project, a wearable device is to be made with the help of flex sensors and an accelerometer sensor which is to be used by a speech disorder person to communicate with others. There are many sign language systems available, considering the American Sign Language (ASL) system for each alphabet and each numerical dataset is to be created. A comparative study is to be made by applying different machine learning algorithms for the created dataset. The ML algorithm with better performance is identified and to be deployed in microcontroller in the future.