Deep learning-based approach for identifying writers' gender using Sinhala handwritten text

Handwriting analysis plays a significant role in forensic science, psychological profiling, and document authentication due to the unique patterns in each person’s handwriting, influenced by motor skills and habits. Building on this, Artificial Intelligence and machine learning techniques have been widely applied to improve handwriting analysis in popular languages such as English and Arabic. However, similar research on Sinhala is limited due to its complex script, which includes a large number of characters and many combined forms. This study developed a model that accurately predicts a writer’s gender using Sinhala handwritten text. To achieve this, a diverse dataset was collected from individuals aged 15 to 70. Research combines traditional handwriting features such as skew angle, irregularity, and letter size with deep learning methods to improve prediction accuracy. In this study, KNN, SVM, RF, and DNN models were tested, and out of the models, the Deep Neural Network (DNN) was the best performer in classifying gender. This work contributes to forensic science and psychological research. Further, it lays a foundation for future studies on lesser-studied languages.