Deep Arabic document layout analysis

A Review of Arabic Document Analysis Methods

“…Furthermore, the remarked image classifier CNN [40] (convolutional neural network) is also used in recent studies, where Amer et al [11] proposed two CNN classifiers based on patches and zones for segmenting the text and non-text regions in newspapers. Magazines and newspaper pages are considered the most complex structures that can exist due to the dissimilarity, variety, and comprehensiveness that cannot be found in other types.…”

Deep Arabic Font Family and Font Size Recognition

“…In [6] we proposed a method for Arabic document layout analysis for text localization and separation from images in Arabic newspapers. The method mainly uses a deep convolutional neural network (D-CNN) to classify zones/patches of the document image as text or non-text then the text lines and words of textual zones are extracted.…”

Developing an Automatic Layout Analysis System for Ottoman Population Registers

“…Amer ve dig. [11] Arapça gazeteler ve Arapça basılı metinler için CNN tabanlı bir belge düzeni analiz sistemi önerdi. Metin ve metin olmayan bölgeleri bulmada yaklaşık % 90 dogruluk elde ettiler.…”