| The handwriting recognition field has preoccupied the scientific community for several years. The complexity encountered in this sector is due to the fact that each individual has a unique way of writing. Various methods have been examined and evaluated throughout the years, with the sole purpose of achieving sustainable results. The introduction of neural networks, specifically the use of convolutional neural networks (CNN) and recurrent neural networks (RNN), has presented dependable results in the handwriting recognition field. In this paper, we introduce a model that recognizes handwritten words without pre-segmenting the words into characters. The model consists of a CNN for the extraction of features, a RNN for the prediction procedure and a final layer (CTC) for decoding the prediction. We conducted many experiments on the well-known IAM handwriting database and attained an accuracy of 77.22% and a character error rate (CER) of 10.4%. |
*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.