RWTH OCR: A Large Vocabulary Optical Character Recognition System for Arabic Scripts

Dreuw, Philippe; Rybach, David; Heigold, Georg; Ney, Hermann

doi:10.1007/978-1-4471-4072-6_9

Cited by 24 publications

(5 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In what follows, we detail the most important points that prove the superiority and robustness of BoF-deep SAE-HMM system compared to the top proposed systems. The system proposed by [50] used a writer adaptive training to train writer dependent models. Thanks to the robustness of the extracted features based on the deep SAE, BoF-deep SAE-HMM system fits effectively the variety of writers.…”

Section: Comparison With State-of-art Systemsmentioning

confidence: 99%

Deep Sparse Auto-Encoder Features Learning for Arabic Text Recognition

et al. 2021

View full text Add to dashboard Cite

One of the most recent challenging issues of pattern recognition and artificial intelligence is Arabic text recognition. This research topic is still a pervasive and unaddressed research field, because of several factors. Complications arise due to the cursive nature of the Arabic writing, character similarities, unlimited vocabulary, use of multi-size and mixed-fonts, etc. To handle these challenges, an automatic Arabic text recognition requires building a robust system by computing discriminative features and applying a rigorous classifier together to achieve an improved performance. In this work, we introduce a new deep learning based system that recognizes Arabic text contained in images. We propose a novel hybrid network, combining a Bag-of-Feature (BoF) framework for feature extraction based on a deep Sparse Auto-Encoder (SAE), and Hidden Markov Models (HMMs), for sequence recognition. Our proposed system, termed BoF-deep SAE-HMM, is tested on four datasets, namely the printed Arabic line images Printed KHATT (P-KHATT), the benchmark printed word images Arabic Printed Text Image (APTI), the benchmark handwritten Arabic word images IFN/ENIT, and the benchmark handwritten digits images Modified National Institute of Standards and Technology (MNIST).

show abstract

Section: Comparison With State-of-art Systemsmentioning

confidence: 99%

Deep Sparse Auto-Encoder Features Learning for Arabic Text Recognition

et al. 2021

View full text Add to dashboard Cite

show abstract

“…For instance the Georgia Tech gesture toolkit GT 2 K [41] was created based on a popular speech recognition toolkit known as HTK to provide tools that support gesture recognition research. Additionally, although RASR toolkit was originally developed for speech recognition, it has proved to be flexible and could be easily adapted for different applications such as SLR [42] [23] and optical character recognition [43]. An example of a toolkit created specifically for gesture recognition is the gesture recognition toolkit GRT [44] created by Gillian and Paradiso in 2014 with emphasis on real time recognition.…”

Section: B Hidden Markov Models (Hmms)mentioning

confidence: 99%

Multiple Proposals for Continuous Arabic Sign Language Recognition

2019

View full text Add to dashboard Cite

The deaf community relies on sign language as the primary means of communication. For the millions of people around the world who suffer from hearing loss, interaction with hearing people is quite difficult. The main objective of Sign language recognition (SLR) is the development of automatic SLR systems to facilitate communication with the deaf community. Arabic SLR (ArSLR) specifically did not receive much attention until recent years. This work presents a comprehensive comparison between two different recognition techniques for continuous ArSLR, namely a Modified k-Nearest Neighbor (MKNN) which is suitable for sequential data and Hidden Markov Models (HMMs) techniques based on two different toolkits. Additionally, in this work, two new ArSL datasets composed of 40 Arabic sentences are collected using Polhemus G4 motion tracker and a camera. An existing glove-based dataset is employed in this work as well. The three datasets are made publicly available to the research community. The advantages and disadvantages of each data acquisition approach and classification technique are discussed in this paper. In the experimental results section, it is shown that classification accuracy for sign sentences acquired using a motion tracker are very similar the classification accuracy for sentences acquired using sensor gloves. The modified KNN solution is inferior to HMMs in terms of the computational time required for classification.

show abstract

“…The recognition system is not fully tuned with simple preprocessing and basic training of the classifier. We use a Hidden Markov Model (HMM) system based on the RWTH OCR [7]. The first step in any pattern recognition system is the data preparation (or preprocessing) and the feature extraction.…”

Section: System Descriptionmentioning

confidence: 99%

Open Vocabulary Arabic Handwriting Recognition Using Morphological Decomposition

Hamdani

Mousa

Ney

2013

2013 12th International Conference on Document Analysis and Recognition

Self Cite

View full text Add to dashboard Cite

Abstract-The use of Language Models (LMs) is a very important component in large and open vocabulary recognition systems. This paper presents an open-vocabulary approach for Arabic handwriting recognition. The proposed approach makes use of Arabic word decomposition based on morphological analysis. The vocabulary is a combination of words and subwords obtained by the decomposition process. Out Of Vocabulary (OOV) words can be recognized by combining different elements from the lexicon. The recognition system is based on Hidden Markov Models (HMMs) with position and context dependent character models. An n-gram LM trained on the decomposed text is used along with the HMMs during the search. The approach is evaluated using two Arabic handwriting datasets. The open vocabulary approach leads to a significant improvement in the system performance. Two different types experiments for two Arabic handwriting recognition tasks are conducted in this work. The proposed approach for open vocabulary allows to have an absolute improvement of up to 1% in the Word Error Rate (WER) for the constrained task and to keep the same performance of the baseline system for the unconstrained one.

show abstract

RWTH OCR: A Large Vocabulary Optical Character Recognition System for Arabic Scripts

Cited by 24 publications

References 33 publications

Deep Sparse Auto-Encoder Features Learning for Arabic Text Recognition

Deep Sparse Auto-Encoder Features Learning for Arabic Text Recognition

Multiple Proposals for Continuous Arabic Sign Language Recognition

Open Vocabulary Arabic Handwriting Recognition Using Morphological Decomposition

Contact Info

Product

Resources

About