Handwritten Indic Script Recognition Using Deep Neural Network

Automatic script recognition has multiple applications, such as document searching, author identification, document archiving, and optical character recognition. Online script recognition operates on digital copies of papers, while offline script recognition involves scanning printed or handwritten materials before initiating the recognition process. However, identifying and recognizing various script types within a single page poses a challenge for automatic script recognition systems functioning in multilingual environments. Numerous published methods exist, primarily relying on traditional machine learning approaches. However, deep learning techniques have gained popularity due to their enhanced optimization and efficiency. Consequently, this paper employs a customized deep neural network for automatic script recognition of handwritten Indian scripts. The proposed method undergoes testing on the benchmark PHD_Indic_11 dataset, which comprises word images written in 11 Indian languages.