AI / Computer Vision
Real-Time Indian Sign Language Recognition
Developed a comprehensive real-time Indian Sign Language recognition system using YOLO and YOLO-NAS architectures. The system supports both single and multi-person gesture recognition with sentence-level understanding. Integrated MediaPipe for skeletal keypoint extraction, significantly enhancing gesture accuracy.
YOLO-NASMediaPipePythonTensorFlowOpenCVNLPLSTM
GitHub Repository
IEEE ICONAT 2024
🎯
The Problem
Deaf and hard-of-hearing individuals face significant communication barriers. Existing sign language recognition systems lack real-time capability and multi-person support.
💡
The Solution
Built a bidirectional translation system between text, voice, and ISL using deep learning and NLP, achieving 15% improved accuracy and 75ms reduced latency.
// OUTCOMES
Key Results
15% increase in sign recognition accuracy
75ms reduction in translation latency
Support for multi-person gesture recognition
Sentence-level sign language understanding