Image and video data analysis has become an increasingly important research area with applications in different domains such as security surveillance, healthcare, augmented virtual reality, image editing, activity recognition, synthetic content generation, distance education, telepresence, remote sensing, sports analytics, art, non-photorealistic rendering, search engines, social media. Recent advances Artificial Intelligence (AI) particularly deep learning have sparked new challenges led to...

The recognition of Arabic script and its derivatives such as Urdu, Persian, Pashto etc. is a difficult task due to complexity this script. Particularly, Urdu text more Nasta'liq writing style. style inherits complex calligraphic nature, which presents major issues owing diagonality in writing, high cursiveness, context sensitivity overlapping characters. Therefore, the work done for cannot be directly applied recognition. We present Multi-dimensional Long Short Term Memory (MDLSTM) Recurrent...

Tables are an easy way to represent information in a structural form. Table recognition is important for the extraction of such from document images. Usually, modern OCR systems provide textual coming tables without recognizing actual table structure. However, structure get contextual meaning contents. heterogeneous documents challenging due variety layouts. It becomes harder where no physical rulings present table. This work proposes novel learning based methodology contents Textual...

OCR of multi-font Arabic text is difficult due to large variations in character shapes from one font another. It becomes even more challenging if the rendered at very low resolution. This paper describes a multi-font, resolution, and open vocabulary system based on multidimensional recurrent neural network architecture. For this work, we have developed various systems, trained for single-font/single-size, single-font/multi-size, multi-font/multi-size data well known printed image database...

This paper presents the first Pashto text image database for scientific research and thereby dataset with complete handwritten printed line images which ultimately covers all alphabets of Arabic Persian languages. Language like Pashto, written in a complex way by calligraphers, still requires mature Optical Character Recognition (OCR), system. Although 50 million people use this language both oral communication, there is no significant effort devoted to recognition Script. A real 17,015...

Background: The development of Pancasila character in early childhood is very important, especially the context working families who often face challenges providing optimal education. includes values such as justice, tolerance, and mutual cooperation, which need to be instilled from an age form a generation with integrity. Research Objectives: This research aims explore strategy developing families, well understand experiences faced by parents process. Method: study uses qualitative...

Optical Character Recognition (OCR) of cursive scripts like Pashto and Urdu is difficult due the presence complex ligatures connected writing styles. In this paper, we evaluate compare different approaches for recognition such ligatures. The include Hidden Markov Model (HMM), Long Short Term Memory (LSTM) network Scale Invariant Feature Transform (SIFT). Current state art in script assumes constant scale without any rotation, while real world data contain rotation variations. This research...

Optical character recognition (OCR) of machine printed Latin script documents is ubiquitously claimed as a solved problem. However, error free OCR degraded or noisy text still challenging for modern systems. Most recent approaches perform segmentation based recognition. This tricky because itself problematic. paper describes line approach using multi layer perceptron (MLP) and hidden markov models (HMMs). A scanning neural network-trained with level contextual information special garbage...

This paper presents a deep learning benchmark on complex dataset known as KFUPM Handwritten Arabic TexT (KHATT). The KHATT data-set consists of patterns handwritten text-lines. contributes mainly in three aspects i.e., (1) pre-processing, (2) based approach, and (3) data-augmentation. pre-processing step includes pruning white extra spaces plus de-skewing the skewed We deploy approach Multi-Dimensional Long Short-Term Memory (MDLSTM) networks Connectionist Temporal Classification (CTC)....

This work presents state-of-the-art results on one of the complex datasets; known as KHATT. The KHATT dataset shows patterns for Arabic handwritten text. We have achieved better performance in terms Character Recognition by implementing most successful deep learning approach based Long Short-Term Memory (LSTM) networks. Connectionist Temporal Classification (CTC) is used a final layer to align predicted labels according probable path. application MDLSTM scans text-lines all direction cover...

Segmentation and recognition of screen rendered text is a challenging task due to its low resolution (72 or 96 ppi) use antialiased rendering. This paper evaluates Hidden Markov Model (HMM) techniques for OCR -- both on isolated characters text-lines compares it with the performance other commercial open source systems. Results show that HMM-based methods reach yield above 98% character level accuracies characters.

Atomic segmentation of cursive scripts into constituent characters is one the most challenging problems in pattern recognition. To avoid script, concrete shapes are considered as recognizable units. Therefore, objective this work to find out alternate units Pashto script. These alternatives ligatures and primary ligatures. However, we need sound statistical analysis appropriate numbers In work, a corpus 2, 313, 736 words extracted from large scale diversified web sources, total 19, 268...

Many languages use Arabic script for written communication either in basic or augmented form. These include Urdu, Pashto, Persian, etc. As the primary characters are shared among all these languages, it is possible to take advantage of visual similarities Optical Character Recognition (OCR). OCR models optimized individual have been proposed. However, best our knowledge, there no attempt develop a single system more than one language. The contributions presented work are: First, investigates...

This dossier is aimed at individuals involved in the development, implementation, and application of AI-supported systems field education, particularly those who are dealing with metadata this context. The insights recommendations described based on interviews five experts from USA UK, research standards, our personal experiences context INVITE competition. provides an overview current development learning existing standards. It shows efforts obstacles to introducing standards international...

Orientation detection is an important preprocessing step for accurate recognition of text from document images. Many existing orientation techniques are based on the fact that in Roman script ascenders occur more likely than descenders, but this approach not applicable to other scripts like Urdu, Arabic, etc. In paper, we propose a discriminative learning Urdu documents with varying layouts and fonts. The main advantage our it can be applied easily accurately. Our classification individual...

Document script recognition is one of the important preprocessing steps in a multilingual optical character (MOCR) system. A MOCR system requires prior knowledge to accurately recognize text single document. In documents two scripts can be mixed together within line. Many existing methods lack ability multiple Besides, these usually use dependent features for thereby limiting their scope particularly that script. this paper we propose discriminative learning approach multi-script at...

Current approaches for text line segmentation often are either very specialized to specific domains or they depend on many parameters. More specifically, the extraction of text-lines with large sizes, i.e., headings and titles in Arabic like script could not be segmented correctly by state-of-the-art methods. In this work, we present a simple robust text-line approach. The proposed method is tested real Pashto scanned images it outperforms recent independent state art respect performance time.

This paper investigates and analyses the nature of errors occurring in Optical Character Recognition (OCR) for Arabic-like scripts. Existing research on area OCR scripts often focuses achieving best performance terms character error rates. Only little effort targets at analysis these (anomalies) that may occur. One such important anomaly is Space Anomaly. due to presence breaker characters are an essential part The spaces introduced by not depicted ground truth making it hard generalize....

Graph convolution network (GCN) is an important method recently developed for few-shot learning. The adjacency matrix in GCN models constructed based on graph node features to represent the relationships, according which, achieves message-passing inference. Therefore, representation ability of factor affecting learning performance GCN. This paper proposes improved model with feature optimization using cross attention, named GCN-NFO. Leveraging attention mechanism associate image support set...

Cursive handwriting recognition is still a hot topic of research, especially for non-Latin scripts. One the techniques which yields best results based on recurrent neural networks: with neurons modeled by long short-term memory (LSTM) cells, and alignment label sequence to output performed connectionist temporal classification (CTC) layer. However, network training time consuming, unstable, tends over-adaptation. reasons bootstrap process, aligns data more or less randomly in early...