Real-Time Sign Language Interpretation System

This work focuses on designing a real time sign language interpretation system that may recognize different signs in real time and interpret the information in text form to the users. Dynamic hand gesture recognition is performed by applying image classification techniques. The model takes sign language images from the webcam as input, process them and convert them into respective written textual formats. For this purpose, TensorFlow Object Detection API is used and transfer learning mechanism is applied using SSD MobileNet. The presented model is tested on varied samples presented by different persons and the results obtained in all the experiments performed shows the effectiveness of the presented model in correctly interpreting the sign language words from image to respective text format with an average accuracy of 95%.