EDUSIGN
EduSign combines educational videos with AI-powered sign recognition to create an accessible, age-appropriate learning platform for children ages 2-15, through interactive tutorials and real-time web.
Created on 22nd March 2025
•
EDUSIGN
EduSign combines educational videos with AI-powered sign recognition to create an accessible, age-appropriate learning platform for children ages 2-15, through interactive tutorials and real-time web.
The problem EDUSIGN solves
EduSign: Making Sign Language Learning Accessible
Empowering Communication Through Innovation
EduSign transforms sign language education by providing:
Accessible Learning for Families - Parents of deaf or hard-of-hearing children can learn sign language alongside their kids, strengthening family communication without expensive private tutors.
Educational Support for Schools - Teachers gain a supplementary tool to support inclusive classrooms where deaf and hearing students learn together.
Self-Paced Progress Tracking - Learners receive immediate feedback on sign accuracy, eliminating the uncertainty of self-study and accelerating skill development.
Age-Appropriate Learning Pathways - Content tailored to developmental stages ensures children ages 2-15 engage with material that matches their cognitive abilities.
Bridging Communication Barriers - Hearing children can learn to communicate with deaf peers, promoting inclusion and reducing social isolation in educational settings.
Remote Learning Solution - Families in areas without access to sign language instructors can receive quality instruction and feedback from anywhere.
By combining structured video tutorials with AI-powered gesture recognition, EduSign makes what was once a challenging learning process more approachable, effective, and engaging for children at all developmental stages.
Challenges we ran into
Overcoming the Webcam-to-AI Integration Challenge
One of the most significant hurdles we faced was reliably capturing and processing webcam frames for real-time sign language recognition during quizzes.
The Problem
When implementing the sign recognition feature, we encountered a critical issue: large latency when transmitting webcam data to our Python AI backend. The initial implementation resulted in:
Slow response times (3-5 seconds delay)
Frequent timeout errors with larger frame sets
High memory consumption crashing the browser
Our Solution
We implemented a multi-faceted approach to resolve this:
Reduced image quality and size: We added preprocessing on the frontend that scaled down images before sending them to the server, significantly reducing payload size.
Limited frame capture rate: Instead of capturing at 30fps, we reduced to 3fps and capped the total number of frames sent to the server for each sign.
Implemented streaming buffer limits in Express to handle larger payloads, preventing server-side memory issues during processing.
Optimized the MediaPipe processing pipeline in our Python backend to prioritize speed over absolute precision.
These changes reduced our response time to under 1 second and improved reliability from 70% to 98%, creating a much more natural and engaging quiz experience for our users.
Tracks Applied (1)
Ethereum Track
ETHIndia