Handwritten Word Recognition in English Language

This work focuses on a comprehensive review of Handwritten word Recognition in the English language. This work is based on improving handwritten word recognition accuracy using deep learning using Convolutional Neural Networks. The dataset used in this project is derived from Kaggle Dataset, comprising of 374,000 of text images with 28 × 28 dimensions. The dataset was divided in the ratio of 80:20 for training and testing, respectively. The models like Transfer Learning model, ResNet, KNN, SVM, and CNN model were used and the accuracy was compared after feature extraction, the training model was predicted initially on test image and then real-time processing is also done on the external images to remove skewness and shadows and finally the desired result with an accuracy of 99.96 and F1 score of 0.9992750 were obtained.