Recognition of Indian Sign Languages

Abstract

The focal point of this project is about  developing a user interface to aid people who have hearing  and speaking disabilities. This application would help in  recognizing the different signs with respect to the single  user. The real-time image is captured using a webcam and  dynamically stored. The elements are initialized and saved  in KN Neighbor classifiers. These classifiers are used to  load the KN Neighbor model. This model predicts the signs  dynamically and is built using machine learning  techniques. Mobile Net is used as a machine learning  package. This model was implemented in three phases. The  first phase deals with the user interface where user images  are captured using a webcam. In the second phase, the  initialized elements are used by the KN Neighbor classifier  and then stored as the KN Neighbor model. The third phase  involves the prediction of signs.