Web-Based Real-Time Gesture Recognition with Voice

WebPredict.ai is a model web developed for converting image into text, followed by the conversion of the predicted text to speech. Images used in the proposed application are taken in the form of hand gestural inputs with the help of webcam. This application also helps to interact with google home or smart home of google. As the gestures captured by the webcam can be given for functionality which works for google home and the application can speak out the gesture inputs with different OS voices. These different OS voices are supported by user's browser and Device. This application was developed on HTML, JavaScript and with the help of Tensorflow.js. Tensorflow.js is an open-source library which is used to train, define, and run machine learning models entirely in the browser, with the help of JavaScript and some high-level API's. The runtime environment for the proposed model is provided by Nodejs which helps in the development of web application and is written in JavaScript. In the proposed application, ML algorithm is running in the web browser, hence there is no need to install any driver or library. Users just need to open the webpage and the program is ready to run. Users can also open the webpage from a mobile device.