Modi Script Recognition Using Convolutional Neural Networks and VGG16: A Deep Learning Approach for Historical Script Analysis

Recognizing historical scripts is a challenging task that requires specialized techniques due to the unique characteristics and variations present in these scripts. The Modi script, a historic script used to write the Marathi language in India, is presented in this research as a deep learning solution for the recognition of the script utilizing convolutional neural networks (CNNs) and the VGG16 architecture. The proposed approach consists of several steps. First, a diverse dataset of Modi script samples is collected, encompassing various characters, words, and sentences written in the script. The dataset is preprocessed to ensure consistency and enhance the model's robustness, including resizing, grayscale conversion, normalization, and data augmentation techniques. After that, the dataset is partitioned into three sets: training, validation, and testing. The VGG16 architecture is employed as the base model for feature extraction, leveraging its ability to learn low-level and high-level features from images. By obtaining handwritten examples from the IEEE Dataport collection, the dataset (training images) for the characters is updated. The model is trained using the training dataset, and hyperparameter tuning is performed using the validation set to optimize its performance. The testing dataset is used to measure the trained model's performance in identifying the Modi script by assessing metrics including accuracy, precision, recall, and F1 score. The performance of the model is assessed, and any updates or modifications that are required to increase the model's accuracy are implemented. The developed model can be deployed for prediction on unseen Modi script images, enabling the recognition of characters or words within the script. This approach provides a promising solution for historical script analysis and contributes to the preservation and understanding of the Modi script's cultural and linguistic heritage.