SignLink: A Real-Time Web-based Sign Language Recognition System using MediaPipe and PeerJS

Sign language is an essential communication for individuals with hearing and speech impairments, yet it remains inaccessible to much of the general population. This paper presents SignLink, a real-time web-based system that translates hand gestures into corresponding alphabetic characters using MediaPipe Hands and PeerJS. The application captures live video input, detects and tracks 21 hand landmarks, and computes angular relationships between finger joints to classify gestures representing letters and words. PeerJS enables peer-to-peer video communication, allowing users to join or create virtual rooms for interactive sign recognition sessions. The system operates directly within a browser, eliminating the need for additional installations or hardware. Experimental testing demonstrates the potential of the framework to provide accurate, low-latency gesture recognition while promoting accessibility and inclusivity. SignLink highlights the integration of computer vision and real-time web technologies for effective human–computer communication.