Arabic Script based Digit Recognition Systems

Naz, Saeeda; Ahmed, Saad Bin; Ahmad, Riaz; Razza, Muhammad

doi:10.2991/racs-15.2016.11

Cited by 8 publications

(2 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our this work is focused on text classification and for classification tasks most commonly used evaluation metrics are F-score and Accuracy [22,23,1].…”

Section: Evaluation Measurementioning

confidence: 99%

Large Scale Font Independent Urdu Text Recognition System

Rehman¹,

Hussain²

2020

Preprint

View full text Add to dashboard Cite

OCR algorithms have received a significant improvement in performance recently, mainly due to the increase in capabilities of artificial intelligence algorithms. However this advancement is not evenly distributed over all languages. Urdu is among the languages which did not receive much attention, specially in the font independent perspective. There exists no automated system that can reliably recognize printed Urdu text in images and videos across different fonts. To help bridge this gap, we have developed Qaida, a large scale data set with 256 fonts and complete Urdu lexicon. We have also developed Convolutional Neural Network (CNN) based classification models which can recognise Urdu ligatures with 84.2% accuracy. Moreover we demonstrate that our recognition network can not only recognise the text in the fonts it is trained on, but can also reliably recognise text in unseen (new) fonts. To this end, this paper makes following contributions: (i) we introduce a large scale, multiple (more than 200) fonts based data set for printed Urdu text recognition 1 ; (ii) we have designed, trained and evaluated a CNN based model for Urdu text recognition; (iii) we experiment with incremental learning methods to produce state-of-the-art results for Urdu text recognition. All the experiment choices were thoroughly validated via detailed empirical analysis. We believe that our this study can serve as the basis for further improvement in the performance of Urdu OCR systems.

show abstract

“…Our this work is focused on text classification and for classification tasks most commonly used evaluation metrics are F-score and Accuracy [22,23,1].…”

Section: Evaluation Measurementioning

confidence: 99%

Large Scale Font Independent Urdu Text Recognition System

Rehman¹,

Hussain²

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Dozens of viable approaches can be developed and used for character and digit recognition purposes in different languages [1], [4], [23]. A number of well-known techniques, which were introduced for other machine learning and pattern recognition applications, may also be used in a diversity of character and digit recognition contexts [1], [2], [24]. Artificial Neural Network (ANN) as an example, has been excessively applied in different topologies for learning to recognize and classify characters and numerals [23], such as Multilayer [7], [18], [21], Probabilistic [3], Convolutional [6], [19], [25], and Back Propagation [9], [10], [14], [20] Neural Networks.…”

Section: Introductionmentioning

confidence: 99%

Efficient Gabor-Based Recognition for Handwritten Arabic-Indic Digits

Jaha¹

2019

ijacsa

View full text Add to dashboard Cite

In daily life, the need of automatically digitizing paper documentations and recognizing textual images is still present with existing and potential upcoming rooms for improvements, especially for languages like Arabic, which is unlike English as an instance, has more complex context and not been extensively supported by research in a such domain. As yet, the available online offline optical character recognition (OCR) systems have utilized functional techniques and achieved high performance mainly on machine printed data images. However, in case of handwritten script, the recognition task becomes highly unconstrained and much more challenging. Amongst a large verity of recognizable multilingual characters, handwritten digit recognition is a considerably useful task for different purposes and countless applications. In this research, the focus is on Arabic (known today as Indic or Indian) digit recognition using different proposed Gabor-based approaches in several combinations with different classification methods. The proposed approaches are trained and tested using 91120 digit samples of two independent standard databases (Arabic-Handwritten-Digits and AHDBase), allowing performance variability assessments and comparisons not only between the different combinations of features and classifiers but also between different datasets. The proposed Arabic-Indic digit recognition system achieves high recognition rates reach up to 99.87%. This research practically shows that one of the proposed approaches with significant dimensionality reduced features remains attaining a high recognition rate with low complexity time, which can be hence recommended further for online digit recognition systems.

show abstract