Age, gender and handedness prediction using handwritten text: A comprehensive survey

Handwriting recognition is widely acknowledged as one of the most challenging tasks in pattern recognition and machine learning. This study explores the intriguing field of age, gender and handedness prediction through handwriting analysis by offering valuable insights into its potential applications in forensics, psychology and education. A comprehensive survey has been conducted encompassing an introduction, motivational aspects, background information, sources of data, analysis schemes, survey protocols, reported findings and methodologies used for predicting age, gender and handedness in both Indic and non-Indic scripts. The investigation highlights research gaps and concludes with an in-depth analysis of unresolved issues accompanied by a roadmap for future advancements in this domain. The survey focuses on a systematic examination of eight Indic scripts (Assamese, Bangla, Devanagari, Gurumukhi, Kannada, Malayalam, Tamil and Telugu) and six non-Indic scripts (Arabic, Chinese, Japanese, Persian, Roman and Thai) and it can be concluded that non-Indic scripts attain more accuracy in comparison with Indic scripts. Furthermore, the study mostly focuses on providing a catalog of publicly accessible online datasets featuring diverse handwriting samples from various scripting languages by giving a roadmap for further research in this area.

Automated summary

This study investigates the field of age, gender, and handedness prediction through handwriting analysis, offering valuable insights into its applications in forensics, psychology, and education. A comprehensive survey is conducted on Indic and non-Indic scripts, highlighting research gaps and providing a roadmap for future advancements. The study concludes that non-Indic scripts achieve higher accuracy compared to Indic scripts, and focuses on providing a catalog of publicly accessible datasets for further research in this area.