A survey on the online Arabic handwriting recognition: challenges, datasets, and future directions
Hasanien Ali Talib Alothman, Wafa Lejmi, Mohamed Ali Mahjoub · 2025
The increasing demand for effective methods to recognize digital handwritten Arabic text has led to the creation of numerous online Arabic manuscript datasets to validate handwriting recognition systems. This tendency has been intensified, especially after the increasing use of devices dependent on digital pen writing, which requires continuous development to achieve an advanced stage of discrimination. This research aims to resolve the confusion problem in choosing a convenient open-source dataset that researchers can use to develop and examine handwriting recognition methods. This paper provides inclusive benchmarking of available and state-of-the-art online Arabic manuscript datasets and their unique features. It also provides a brief overview of the preprocessing and feature extraction methods used in the most important research. Research findings include identifying the most efficient databases and providing recommendations for future use.