الفهرس | Only 14 pages are availabe for public view |
Abstract Recognizing text in images has many useful applications, which include document archiving and searching. Improving the accuracy of Arabic text recognition in imagery requires a big modern dataset for machine-learning models learning. This thesis proposes a new dataset, called QTID, for Quran Text Image Dataset, the first Arabic dataset that includes Arabic diacritics. Experimental evaluation shows that current best Arabic text recognition engines cannot work well with word images from the proposed dataset. Two deep learning models was proposed that learned using QTID. Comparing these models outputs to current best Arabic text recognition engines shows that their accuracy outperforms these engines |