Real-time sign spelling and translation
A multi-user real-time sign-spelling prototype using YOLO and skeletal recognition.
YOLOComputer VisionPython
Overview
A multi-user real-time sign-spelling and translation prototype connected to sign-language research and publications.
Problem
Multi-person scenes and continuous motion complicate real-time recognition and tracking.
Solution
The prototype combines YOLO with skeletal cues in a repeatable real-time recognition pipeline.
Role / Contributions
Research, prototype development, and student-paper authorship.
Core development
For the 2025 E-System Creative Application Competition, I trained the real-time YOLO recognition model and implemented the conversion from recognized gestures to text, forming the core sign-spelling workflow.