Real Time Sign Language Interpreter Using Live Video Feed using Deep Learning

Abstract - This paper presents a real-time sign language interpreter system designed to bridge the communication gap between Deaf/non-verbal individuals and non-signers. The proposed system captures hand gestures through a standard webcam and utilizes a Convolutional Neural Network (CNN) for accurate gesture recognition. Preprocessing steps such as Histogram of Oriented Gradients (HOG) and Gaussian Mixture Models (GMM) are used for background elimination and feature extraction. The recognized gestures are translated into both text and speech, enabling seamless interaction in real-world environments. The system also features a user-friendly interface and a modular web-based platform for future expansion. Unlike traditional glove-based or high-complexity models, this solution is lightweight, cost-effective, and deployable on standard hardware, making it highly accessible for educational, professional, and assistive applications. Key Words: Sign Language Recognition, Deep Learning, Convolutional Neural Network, Real-Time Gesture Detection, Assistive Technology