Complex Odia Handwritten Character Recognition using Deep Learning Model

One of the ongoing uses of offline OCR is the identification of Odia character pictures. Here, an effort has been made to create a useful feature extraction method that will aid in recognizing handwritten digits, basic scripts, and compound characters in Odia. Here, character recognition has been implemented using three different types of techniques. First, Three feature extraction techniques are used separately to identify the different characters. a feature collection is combined with a set of common machine learning methods after that. In the second strategy, well-known RNN and CNN carry out character recognition by offering the same set of features but delaying instantaneous visuals, unlike conventional networks. The final challenge was to include a set of complex characters from the Odia language into our proposed framework for character recognition. On this dataset with 120 classes of characters, the suggested technique obtains a recognition accuracy of 88.76%.