Evaluating Aesthetics of Arabic Handwriting: Hand Motion-Based Approach

Vahan Babushkin, Haneen Alsuradi, Muhamed Osman Al-Khalil, Mohamad Eid

Abstract: Aesthetics pertain to the visual appeal of handwritten text, having significant cultural and artistic importance for all script-based languages. Aesthetics play an important role in Arabic orthography, influencing the overall quality and legibility of Arabic handwriting. Despite the importance of aesthetics, there are only a few studies leveraging machine learning to evaluate handwriting aesthetics, and, to our best knowledge, no studies focusing on automated aesthetics evaluation of Arabic handwriting. A system for automated aesthetics evaluation can eliminate expert bias and assist those passionate about calligraphy in achieving a desired aesthetic level. We propose a Temporal Convolutional Network (TCN)-based system for automated evaluation of Arabic handwriting aesthetics. Unlike existing studies that rely on static text features, our approach uses dynamic data from hand and stylus kinematics, enabling real-time aesthetics assessment and immediate modifications. The proposed model achieves 94% accuracy in classifying aesthetic scores. Additionally, we employ interpretable machine learning techniques to identify the most influential features, offering insights into factors contributing to handwriting aesthetics. The potential applications of proposed model can be found in educational tools, i.e., personalized learning systems, for handwriting difficulties detection, individual authentication, development of remediation, intervention methods, arts, and calligraphy assessment.

External IDs:doi:10.1007/978-3-031-97254-6_7