Letter and Person Recognition in Freeform Air-Writing Using Machine Learning Algorithms

This project presents an Arduino- and PC-based intelligent system for Person and Letter Recognition using Freeform Air-Writing, enhanced with machine learning techniques. The system initially performs person detection through a web camera and becomes active only when a human presence is identified, ensuring intentional interaction. Initially it is in finger gesture mode,Two dedicated push buttons are incorporated to provide flexible input modes: the first button activates air-writing mode, where freeform writing is performed in the air using a trackable object such as an LED, IR marker, or visible pointer; the second button enables emergency alert. The motion of the tracked object or finger gestures is continuously captured and analyzed using computer vision algorithms, converting movement trajectories into letters and words. These recognized writing patterns are further processed using machine learning models to associate unique handwriting characteristics with specific individuals, enabling person recognition. The recognized letters and identified person details are displayed in real time on an LCD or PC interface. By combining human detection, dual-mode input control, motion tracking, and learning-based recognition, the proposed system offers an intuitive, secure, and contactless human–computer interaction solution. The design is portable, cost-effective, and suitable for applications such as smart interfaces, secure authentication, assistive technologies, and next-generation input systems.

Keywords:

• Air-Writing Recognition
• Machine Learning
• YOLO (You Only Look Once)/Mediapipe
• Person Identification
• Arduino UNO