Two students at the International Institute of Information Technology, Hyderabad (IIIT-H), have developed a browser plugin that converts highlighted web text into Indian Sign Language (ISL) through a 3D avatar, offering a low-cost approach to improving online access for people with hearing impairment.
The tool, called SignPlug, was developed by Masthan Sharif Shaik and Mohammed Raza, third-year BTech students at IIIT-H’s Centre for Visual Information Technology. Users can highlight text on a webpage and select a “click to sign” button to generate a signing sequence. The plugin can also be integrated into websites without specialised hardware or server-based artificial intelligence models, according to the report.
The project addresses a gap in common digital accessibility tools. Websites may offer captions, screen readers and high-contrast modes, but generally do not provide content in sign language. Existing signing-avatar systems can involve motion-capture equipment, computationally intensive AI models or sign-notation systems that require specialist linguistic input, the developers told The Times of India.
For SignPlug, the team used videos from the official Indian Sign Language dictionary of the Indian Sign Language Research and Training Centre (ISLRTC). Computer vision was used to reconstruct the signer’s body, hand and finger movements in three dimensions. Because automated reconstruction can produce inaccurate finger positions and alter the meaning of a sign, the students developed an annotation tool called Gizmo. It enables them to correct crucial poses without manually editing every frame.
The system also uses an offline pre-processing step to convert sentences into ISL word order. A sentence such as “He will come home” is rearranged as “He home come” before the relevant stored motions are retrieved and blended into a continuous signing sequence. Words without corresponding signs are represented letter by letter through finger movements.
The research paper, titled “Sign Plug – ISL plugin for the web”, has been accepted for the 17th Indian Conference on Computer Vision, Graphics and Image Processing, scheduled to take place in Kolkata in December 2026. Reported tests with deaf and hard-of-hearing signers found that the avatar’s signs were understood in about 89% of sentences.
That result also highlights the technology’s current limits. The avatar does not yet include facial features, even though facial expressions are an important part of ISL grammar and some signs change according to context. The developers plan to add facial expressions and improve contextual signing in a future version.
The project demonstrates how browser-based tools could supplement existing digital accessibility measures without requiring users or website operators to install specialised equipment. However, the report does not establish how widely SignPlug is currently deployed, whether it has been adopted by public-service websites, or how its performance compares with interpretation by human sign-language users. Those questions will be important in assessing whether the research can move from a student-developed prototype to a dependable accessibility layer for India’s digital services.


