Hybrid Handwritten Devanagari Word Recognition System

  • Shalaka Deore, A. Pravin, Aditi Banait, Ravindra Karande, Neha Lokre, Supriya Chougule


Various languages such as Marathi, Hindi, Sanskrit, Kashmiri, Bhojpuri etc., belong to a set of languages called the Indo-Aryan languages. These languages have the common base of Devanagari script. Handwritten word recognition has played an important role in many recent applications like reading postal address, bank check amounts, forms. Handwriting recognition of Devanagari script proves very challenging due variations in rendering and the cursive nature of the handwriting. Recognition of Devanagari word consists of image correction, feature extraction, classification and character recognition. In Image correction, RGB images are converted into Grey-Scale images. Feature extraction is a significant step in recognition system. Among various feature extraction techniques, proposed work uses Sliding Window Approach. The model i.e. Convolutional Neural Network (CNN) then classifies the characters, collectively recognizing the Devanagari word.