Gender Classification from Offline Handwriting Images in Urdu Script: LeNet-5 and Alex-Net

Everyone has a distinct handwriting style, even identical twins who share the same genes. This distinctiveness has allowed researchers to develop systems that can classify the writer's traits, including gender, age, handedness, and others. A gender classification system, for instance, classifies handwriting patterns to predict the writer's gender. With the advancement of deep learning, it is now possible to automatically extract features from the document and classify them. In this paper, two gender classification models based on Alex-Net and LeNet-5 CNN were proposed, trained, and tested on a private Urdu handwriting dataset containing 284,000 pre-segmented characters from 200 males and 200 females. The proposed models achieved state-of- the-art performance compared to existing deep learning models for gender classification. The Alex-Net model achieved an overall accuracy of 99.14%, while the LeNet-5 model achieved an overall accuracy of 98.55%.