Attention-based CNN-ConvLSTM for Handwritten Arabic Word Extraction

Word extraction is one of the most critical steps in handwritten recognition systems. It challenging for many reasons, such as variability writing styles, touching and overlapping characters, skewness problems, diacritics, ascenders, descenders' presence. In this work, we propose a deep-learning-based approach Arabic word extraction. We used an Attention-based CNN-ConvLSTM (Convolutional Long Short-term Memory) followed by CTC (Connectionist Temporal Classification) function. Firstly, text-line input image's essential features are extracted using Convolutional Neural Networks (CNN). The text line's transcription then passed to ConvLSTM learn mapping between them. Finally, alignment images their automatically. tested proposed model on complex dataset known KFUPM Handwritten Text (KHATT \cite{khatt}). consists patterns text-lines. experimental results show apparent efficiency combination, where ended up with success rate 91.7\%.