Deep Learning-Based Sign Language Recognition System

The oldest and natural kind of communication mode in human life is being sign language. Identification is the solitary mode of message for the human being that is mute and hearing impaired. To comprehend them, one has to either learn their language, i.e., sign language. Majority of us do not know how to interpret and sign in return. The object of the mentioned study is to prosper a “sign language recognition in a real time” using image processing and convolutional neural network (CNN) in machine learning technology (MLT). The study deals with an image recognition system that recognizes an unknown test hand images with the known training hand images stored in the database. Hand images are captured through the webcam. The dataset from the webcam (hand images) was first passes through a filter for feature extraction and representation. The submitted data images were then transmitted to a classifier who then predicted the class of hand gestures. This method generates an accuracy of 95.7% for 26 English alphabets. The system proposed in this study aims at tackling this problem to some extent. In this study, I use python programming, Open Computer Vision (OpenCV) library, TensorFlow, and convolutional neural network library.