Website for seamless translation of sign language using KNN and TensorFlow

The Sign Language Translator is a user-friendly website designed to train the sign language that acts as a bridge for the communication between users. The core functionality lies in capturing and training individual signs with their corresponding meanings. It proceeds with a start and stop gesture to maintain a flow of end and beginning of the translation. It uses a minimum of thirty samples to get trained with more accuracy in recognizing the sign language shown. The start and stop gesture is collected and trained. The next step includes the signs that need to be trained with their meaning with atleast thirty samples each. This translator depicts how a machine learning model works by training the samples and testing on the data given by the user. It detects the sign by using the trained samples and gives the result as text with computer generated speech in the background. One can sign multiple words with one gesture and copy the translated text with a click of a button. Its purpose is to allow users to communicate more effectively with their computers and other people. The use of KNN (K-Nearest Neighbours) algorithm contributes to gesture recognition of sign language when trained with samples. Users can initiate and conclude the translation process through start and stop gestures trained by them thus giving a controlled flow. Previous research work in this area has models that can translate trained symbols in 2.3 to 3.1 seconds with a confidence level of 84%. We targeted to focus on this time interval, confidence level and also showcase the project with a straightforward user interface. It also serves more as an educational tool in the view of communication between deaf people and the normal people providing a way to vanish the indifferences. This translator additionally provides the feature of copying the text translated from recognizing the sign language which offers functionalities such as sharing it with others or integrating it to a document. The auditory reinforcement of the translated text is another notable feature of this translator offering a clear vocal representation.