Filters








14 Hits in 4.3 sec

Recognition of Urdu Handwritten Characters Using Convolutional Neural Network

Mujtaba Husnain, Malik Muhammad Saad Missen, Shahzad Mumtaz, Muhammad Zeeshan Jhanidr, Mickaël Coustaty, Muhammad Muzzamil Luqman, Jean-Marc Ogier, Gyu Sang Choi
2019 Applied Sciences  
In this paper, we propose the use of the convolutional neural network to recognize the multifont offline Urdu handwritten characters in an unconstrained environment.  ...  In a very recent work [56] , the authors presented a simple and robust line segmentation algorithm for Urdu handwritten and printed text.  ...  Different classifiers such as the hidden Markov model (HMM), fuzzy logic, the k-nearest neighbor (KNN), hybrid fuzzy HMM, hybrid KNN fuzzy, and the convolutional neural network (CNN) wee used for the classification  ... 
doi:10.3390/app9132758 fatcat:ymmudtesgjf3bch2fqaoy3cf3i

Subword Recognition in Historical Arabic Documents using C-GRUs

Hanadi Hassen, Somaya Al-Madeed, Ahmed Bouridane
2021 TEM Journal  
More specifically, we introduce a hybrid CNN-GRU model where the shallow convolutional network learns robust feature representations while the GRU layers carry out the sequence modelling and generate the  ...  .  A hybrid CNN-GRU architecture is introduced with shallow convolutional layers extracting robust features from subwords while the GRU layers learn to map the feature sequences to subword class labels  ...  Methods This section presents the details of the proposed subword recognition technique that relies on extraction of robust representations using convolutional layers followed by sequence modelling using  ... 
doi:10.18421/tem104-19 fatcat:v4z4eyksevb4pedqqiozfyatky

Quranic Optical Text Recognition Using Deep Learning Models

Masnizah Mohd, Faizan Qamar, Idris Al-Sheikh, Ramzi Salah
2021 IEEE Access  
A Quranic optical character recognition (OCR) system based on convolutional neural network (CNN) followed by recurrent neural network (RNN) is introduced in this work.  ...  A new Quranic OCR dataset is developed based on the most famous printed version of the Holy Quran (Mushaf Al-Madinah), and a page and line-text image with the corresponding labels is prepared.  ...  They used the MDLSTM network as input layers such that the MDLSTM could learn the raw input image features. They used CTC with SoftMax activation function as the output layer.  ... 
doi:10.1109/access.2021.3064019 fatcat:obe2pevoijdwpos3ejrnwo4afa

Learning Spatial-Semantic Context with Fully Convolutional Recurrent Network for Online Handwritten Chinese Text Recognition

Zecheng Xie, Zenghui Sun, Lianwen Jin, Hao Ni, Terry Lyons
2018 IEEE Transactions on Pattern Analysis and Machine Intelligence  
A multi-spatial-context fully convolutional recurrent network (MC-FCRN) is proposed to exploit the multiple spatial contexts from the signature feature maps and generate a prediction sequence while completely  ...  avoiding the difficult segmentation problem.  ...  [6] , [7] , segmentation-free methods [8] , [9] , [10] with long shortterm memory (LSTM) and multi-dimensional long shortterm memory (MDLSTM), and integrated convolutional neural network (CNN)-LSTM  ... 
doi:10.1109/tpami.2017.2732978 pmid:28767364 fatcat:cwulpgzg5bb7jo44n7kbsxgusu

A limited-size ensemble of homogeneous CNN/LSTMs for high-performance word classification [article]

Mahya Ameryan, Lambert Schomaker
2019 arXiv   pre-print
In this paper, an end-to-end convolutional LSTM Neural Network is used to handle both geometric variation and sequence variability.  ...  The networks have similar architectures (Convolutional Neural Network (CNN): five layers, bidirectional LSTM (BiLSTM): three layers followed by a connectionist temporal classification (CTC) processing  ...  For RNN, a multi-dimensional long-short term memory neural network (MDLSTM) [4] is used.  ... 
arXiv:1912.03223v1 fatcat:vxheolvnmbf5fbd2zv3m2cx4gu

HWRCNet: Handwritten Word Recognition in JPEG Compressed Domain using CNN-BiLSTM Network [article]

Bulla Rajesh, Abhishek Kumar Gupta, Ayush Raj, Mohammed Javed
2022 arXiv   pre-print
The proposed model combines the Convolutional Neural Network (CNN) and Bi-Directional Long Short Term Memory (BiLSTM) based Recurrent Neural Network (RNN).  ...  The handwritten word recognition from images using deep learning is an active research area with promising performance.  ...  So, we rely on a hybrid deep learning model consisting of a Convolutional Neural Network (CNN) and a Recurrent Neural Network (RNN).  ... 
arXiv:2201.00947v2 fatcat:e4ovw2hx6jac7ihnifxugdlory

Whole page recognition of historical handwriting [article]

Hans J.G.A. Dolfing
2020 arXiv   pre-print
We conclude that a whole page inference approach without text localization and segmentation is competitive.  ...  We explore its robustness and accuracy compared to a line-by-line segmented approach based on the IAM, RODRIGO and ScribbleLens corpora, in three languages with handwriting styles spanning 400 years.  ...  In earlier work, IAM words and lines have been classified with CNNs and convolutions [40] , LSTMs [17] and dropout [41] as well as HMM-based [42] and hybrid NN-HMM [43] , [44] systems.  ... 
arXiv:2009.10634v1 fatcat:j5ivuqpp75dv5bkwq4dom4qdge

Learning Spatial-Semantic Context with Fully Convolutional Recurrent Network for Online Handwritten Chinese Text Recognition [article]

Zecheng Xie, Zenghui Sun, Lianwen Jin, Hao Ni, Terry Lyons
2017 arXiv   pre-print
A multi-spatial-context fully convolutional recurrent network (MCFCRN) is proposed to exploit the multiple spatial contexts from the signature feature maps and generate a prediction sequence while completely  ...  avoiding the difficult segmentation problem.  ...  methods [8] - [11] with long short-term memory (LSTM) and multi-dimensional long shortterm memory (MDLSTM), and integrated convolutional neural network (CNN)-LSTM methods [12] - [15] .  ... 
arXiv:1610.02616v2 fatcat:qbx7rjazhvhezps43qdg3gzoa4

Continuous Offline Handwriting Recognition using Deep Learning Models [article]

Jorge Sueiras
2021 arXiv   pre-print
For this purpose, we have proposed a new recognition model based on integrating two types of deep learning architectures: convolutional neural networks (CNN) and sequence-to-sequence (seq2seq) models,  ...  Additionally, extensive experimentation of the proposed model for the continuous problem has been carried out to determine its robustness to changes in parameterization.  ...  Models using convolutional networks do not have this problem.  ... 
arXiv:2112.13328v1 fatcat:xkcdw7c2rngd7jsaixsfosqzc4

Robust text line detection in historical documents: learning and evaluation methods

Mélodie Boillet, Christopher Kermorvant, Thierry Paquet
2022 International Journal on Document Analysis and Recognition  
that can correctly segment diverse unseen pages.  ...  Text line segmentation is one of the key steps in historical document understanding.  ...  Doc-UFCN Doc-UFCN is a U-shaped Fully Convolutional Network.  ... 
doi:10.1007/s10032-022-00395-7 fatcat:mdkx6yowmnaavcvsb6yv3aptgy

A limited-size ensemble of homogeneous CNN/LSTMs for high-performance word classification

Mahya Ameryan, Lambert Schomaker
2021 Neural computing & applications (Print)  
In this paper, an end-to-end convolutional LSTM neural network is used to handle both geometric variation and sequence variability.  ...  The networks have similar architectures (convolutional neural network (CNN): five layers, bidirectional LSTM (BiLSTM): three layers followed by a connectionist temporal classification (CTC) processing  ...  [46] obtained very high accuracy for Chinese handwritten character recognition using deep convolutional neural networks and a hybrid serial-parallel ensemble strategy which tries to find an 'expert'  ... 
doi:10.1007/s00521-020-05612-0 fatcat:ceq56kndoreade3pyzz2alr3fa

Urdu Ligature Recognition System: An Evolutionary Approach

Naila Habib Khan, Awais Adnan, Abdul Waheed, Mahdi Zareei, Abdallah Aldosary, Ehab Mahmoud Mohamed
2021 Computers Materials & Continua  
The pre-processing stage removes noise from the sentence images, whereas, in segmentation, the sentences are segmented into ligature components.  ...  The genetic algorithm performs an optimization mechanism using multi-level sorting of the clustered data for improving the classification rules used for recognition of Urdu ligatures.  ...  A system was developed by [14] that used a Convolution Neural Network (CNN) for automatic feature extraction, an MDLSTM was used for classification and recognition.  ... 
doi:10.32604/cmc.2020.013715 fatcat:w7bgzlatbjeu3efblcn2wiyoea

Handwritten text recognition in historical documents

Harald Scheidl, Robert Sablatnig, Stefan Fiel
2018
Convolutional Neural Networks (CNNs) lernen Faltungsmatrizen zum Extrahieren relevanter Bildmerkmale.  ...  The classifier has Convolutional Neural Network (CNN) layers to extract features from the input image and Recurrent Neural Network (RNN) layers to propagate information through the image.  ...  Document analysis methods to detect the text on a page or to segment a page into lines are not discussed.  ... 
doi:10.34726/hss.2018.43931 fatcat:vg6hj72etzbslfkacvu463zlve

Functional Representation of Prototypes in LVQ and Relevance Learning [chapter]

Friedrich Melchert, Udo Seiffert, Michael Biehl
2016 Advances in Intelligent Systems and Computing  
The techniques used in this paper allow for use on other (radicalised) communities as well as building towards automatically detecting radicalism online.  ...  Acknowledgments The research reported has been performed in the context of the project 'Designing and Understanding Forensic Bayesian Networks with Arguments and Scenarios', funded in the NWO Forensic  ...  techniques, like the Convolution Neural Networks (CNN) and Recurrent Neural Networks (RNN) with Long Short Term Memory (LSTM).  ... 
doi:10.1007/978-3-319-28518-4_28 fatcat:uwxvq6txmrba3ajulmblafgh2a