Deep Convolutional Neural Network-Based Recognition of Air-Writing

Computer Engineering

Air-writing, the act of writing characters or words in the air using hand gestures, presents a novel and intuitive approach to human-computer interaction. As touchless interfaces gain popularity, the ability to accurately recognize air-written text has become increasingly important. This paper proposes a robust and efficient air-writing recognition system using Deep Convolutional Neural Networks (CNNs). The proposed model extracts spatial and temporal features from video sequences of hand movements and classifies them into corresponding characters or words. The system is evaluated using a custom dataset collected from multiple users under various conditions. Results demonstrate that the CNN-based model achieves high accuracy and generalizes well across different writing styles and environments, making it suitable for real-time applications in augmented reality, virtual keyboards, and assistive technologies.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.