VidyaNetra is an inclusive educational platform designed for the blind and visually impaired in India. It enables users to independently access learning materials, take exams without scribes, and even learn braille using spatial and binaural audio. With multiple accessibility modules powered by AI, VidyaNetra aims to revolutionize inclusive education and uplift marginalized communities.
- Introduction
- Demo Video
- Presentation
- Technology Stack
- Key Features
- Real-World Impact
- Future Enhancements
- Contributors
- Made At
- Hugging Face Transformers
- Streamlit
- Google Text-to-Speech (gTTS)
- Ollama
- Qwen1.5-7B LLM
- LangChain
- Cloudinary (media upload & management)
- Docling (PDF parsing & structuring)
- Pillow (image processing)
- Converts digital text (PDFs, etc.) into synthesized audio
- Generates digital Braille formats (e.g.,
.brf) - Makes existing written materials accessible for both auditory and tactile learners
- Teaches Braille character patterns using spatial audio cues
- Maps each Braille dot position to a specific 3D sound location
- Supports intuitive, non-visual learning methods
- Secure login using facial and voice recognition
- Reads questions aloud and accepts spoken responses
- Enables independent, scribe-free exams for the visually impaired
- Users can upload educational videos
- Transcribes content using Whisper
- Interactive Q&A using Retrieval-Augmented Generation (RAG) via voice or text input
- Responses are returned in both audio and text format
- Empowers over 9 million visually impaired individuals in India
- Enables independent access to education and examinations
- Scalable integration with government and NGO programs
- Promotes digital equity and accessibility in mainstream education
Implement a voice-powered AI agent for natural language-based platform control, eliminating the need to remember commands or menus.
Convert diagrams and images into textual representations, extract key insights, and deliver them through immersive binaural audio for a deeper auditory learning experience.
Transform charts (bar graphs, line graphs, etc.) into audio experiences using pitch, volume, and spatial cues to represent data—navigated by voice or simple gestures.
Team Name: Bayes Watch
- Harshvardhan Patil
- Darsh Shah
- Vanshika Gautam
- Parth Panchiwala