Two Decades of Bengali Handwritten Digit Recognition: A Survey

Handwritten Digit Recognition (HDR) is one of the most challenging tasks in domain Optical Character (OCR). Irrespective language, there are some inherent challenges HDR, which mostly arise due to variations writing styles across individuals, medium and environment, inability maintain same strokes while any digit repeatedly, etc. In addition that, structural complexities digits a particular language may lead ambiguous scenarios HDR. Over years, researchers have developed numerous offline online HDR pipelines, where different image processing techniques combined with traditional Machine Learning (ML)-based and/or Deep (DL)-based architectures. Although evidence extensive review studies on exists literature for languages, such as English, Arabic, Indian, Farsi, Chinese, etc., few surveys Bengali (BHDR) can be found, lack comprehensive analysis challenges, underlying recognition process, possible future directions. this paper, characteristics ambiguities handwritten along insight two decades state-of-the-art datasets approaches towards BHDR been analyzed. Furthermore, several real-life application-specific studies, involve BHDR, also discussed detail. This paper will serve compendium interested science behind instigating exploration newer avenues relevant research that further better application areas.