MODI Script Handwritten Characters Recognition with Convolutional Neural Networks and Transfer Learning Algorithm

The Modi script is an ancient writing system used in India for centuries. The script is currently used in some parts of India but is mainly considered historical. Here, we examine text recognition methods to understand this archaic language better. Most revenue records were written in the MODI language, primarily utilized by administrative personnel to maintain their accounts. Recognizing characters in the Modi script is essential for preserving and understanding Indian cultural heritage. Handwritten Modi script character recognition is challenging due to its cursive nature and variability in writing styles. This research paper proposes a method for recognizing handwritten characters in the Modi script using Convolutional Neural Networks (CNNs) and transfer learning algorithms VGG16. This approach utilizes handwritten MODI script characters from the IEEE Dataport dataset. The Vgg16 algorithm performs better than the CNN algorithms for utilizing a handwritten Modi script characters dataset.