Contribution of recurrent connectionist language models in improving LSTM-based Arabic text recognition in videos

详细信息    查看全文

• 作者：Sonia Yousfi^a ; ^{sonia.yousfi@orange.com}Author Vitae ; Sid-Ahmed Berrani^a ; ^{sidahmed.berrani@orange.com}Author Vitae ; Christophe Garcia^b ; ^{christophe.garcia@liris.cnrs.fr}Author Vitae
• 关键词：Connectionist language model ; Arabic video OCR ; Decoding ; RNN ; LSTM ; Convolutional Neural Network
• 刊名：Pattern Recognition
• 出版年：2017
• 出版时间：April 2017
• 年：2017
• 卷：64
• 期：Complete
• 页码：245-254
• 全文大小：964 K
• 卷排序：64

文摘

Different recurrent connectionist language models to improve LSTM-based Arabic text recognition in videos. Efficient joint decoding paradigm using language model and LSTM responses. Additional decoding hyper-parameters, extensively evaluated, that improve recognition results and optimize running time. Significant recognition improvement by integrating connectionist language models that outperform n-grams contribution. Final Arabic OCR system that significantly outperforms commercial OCR engine.