Ensemble System of Double Granularity RNN by Linear Combination
Liang Zhang, Shuguang Huang, Hu Rong-gui · Acta Automatica Sinica · 2011
1 Abstract For o†ine text recognition, an ensemble system of double granularity RNN (recurrent neural network) by linear combination is proposed. Firstly, the unknown image is recog- nized by a word RNN. Secondly, based on the recognition re- sult, connected characters are segmented, and then recognized by a character RNN. Characters 0 posterior probabilities are cal- culated by a table-looking method. Finally, the flnal output is decided by combining both RNNs 0 results. Experiment results based on CAPTCHA (completely automated public Turing test to tell computers and humans apart) recognition and handwrit- ten recognition proved the efiectiveness of this ensemble system.