Arabic Handwriting Recognition Competitions

Abstract

Competitions are a general practice to support the development of new systems. Due to the availability of a database of handwritten words Arabic handwriting recognition systems made a considerable improvement. At first this chapter presents the IFN/ENIT-database which is a standard for Arabic handwritten word recognizer development. The second part includes a presentation of the participating systems and the results achieved at five international competitions from the first one at the International Conference on Document Analysis and Recognition ICDAR 2005 to the competition at the ICDAR 2011. The competitions show a remarkable progress of Arabic handwriting recognition system quality during the seven years. Even though most systems used Hidden Markov Model (HMM) based methods and most times HMM based systems were the winners the overall best result was reached by a Neural Net based technique.