Machine Learning-Based Air Writing: A Literature Review on Techniques, Progress, and Challenges

Abstract

This literature review explores the integration of machine learning (ML) techniques in the development of air writing systems, a cutting-edge technology that enables text input through gesture recognition in mid-air. The paper examines various ML approaches that have been applied to air writing, including deep learning, pattern recognition, and gesture classification models, highlighting their effectiveness in improving accuracy and real-time performance. We discuss key advancements in sensor technologies, such as accelerometers, gyroscopes, and vision-based sensors, which have contributed to the progress of air writing systems. Additionally, the review addresses the challenges associated with air writing, including gesture ambiguity, user variability, environmental factors, and the need for large annotated datasets for training ML models. Finally, the paper outlines the potential future directions of air writing, focusing on enhancing system robustness, expanding applications across diverse fields, and achieving seamless human-computer interaction. This review serves as a valuable resource for researchers and developers aiming to advance the field of air writing using machine learning.