Machine learning for Sign Language Recognition

Abstract

Recent advances in mobile computing, wireless sensing and communication technologies, consumer electronics have modernized our cities and living environments. Buildings, roads, and vehicles are now empowered with a variety of smart sensors and objects that are interconnected via machine-to-machine communication protocols, accessible via the Internet, to form what is known as the Internet of Things (IoT). The power of IoT expands when coupled with Machine Learning, since the later o er techniques that allow analyzing the vast amount of data generated by sensors and actuators. Smart buildings are an appealing example of IoT and machine learning applications o ering higher energy saving and occupants satisfaction through dynamic control. Vocal virtual assistants (e.g., Amazon Alexa, Google Home) are now a central component of the smart house. However, they are not adapted to deaf and mute people who communicate using sign language. E cient alternative communication means inside the house are required to assist the interaction of deaf and hearing-impaired people. The main goal of this thesis is to conceive and realize a solution based on machine learning for sign language recognition that allows the control of a smart home environment through gestures. Keywords: Smart buildings, Machine learning, Sign language, Disabled people, HumanComputer interaction.