Text Lines Research Articles

Multiple improvement strategies for high-density text detection based on DBNet

High-Density text detection is actually challenging task, which deserves more attention on its performance improvement. Based on a typical existing text detection model called DBNet, this paper introduces a new approach in improving the performance of it in text-rich and readability interfered scenarios, which are mostly characterized by dozens of lines of text using different font sizes and text background harder for detecting and reading, the most significant scenario is text detection on the passport information page. The approach proposed in this study includes several steps, namely a baseline setting, a method of dataset enrichment and new loss function aiming at raising text detection for short texts and single characters, such methods has achieved enhanced improvement on text detection especially on short texts: an overall precision reaching 0.7854 and recall reaching 0.8758. Compared with the base model, the experimental results demonstrated the effectiveness of the proposed methods in this study.

Generalized Segmentation Algorithm for Dissimilar Script Languages

Optical Character Recognition is considered one of the fastest methods of data entry. OCR converts the text image representation of x and y coordinates representing pixel information to be converted into text data in a particular language. OCR as a field of pattern recognition and document image understanding, OCR requires a challenging job once a different language text is available on the image. Difference in language script will pose different challenges for OCR which requires entirely different approaches and algorithms. Latin scripts require a different approach whereas the Arabic adopted language scripts require a different approach. In this regard, various solutions have been proposed for different languages. Segmentation is considered one of the important tasks in the process of OCR. A good segmentation will definitely increase the accuracy of an OCR. Segmentation includes the segmentation of text lines from text images which are further divided into words. These segmented words are further divided into characters which are to be recognized. A single segmentation algorithm to segment various scripts of the languages is proposed in this study which checks the script and then segments the text image for the further processing in OCR. The proposed generalized algorithm will check the style, direction and other properties of the script and then adopts the segmentation process to segment text lines, words and characters of the language. The proposed algorithm segments more than ten languages of three scripts and segments for their OCRs. These images can be further processed for feature extraction and classification further. The process of OCR for selected languages will be made easier to recognize. Multiple scripts, languages and images were experimented, and the proposed algorithm successfully segmented 32,833 images of text line, words and character image. The algorithm provides 97% accuracy while segmenting these images and can be extended to further languages as well as scripts.

Exploring whether mental health crisis text conversations that include discussion of firearms differ from those without firearms

BackgroundFirearm violence represents a public health crisis in the United States. Yet, there is limited knowledge about how firearms are discussed in the context of mental health emergencies representing a major gap in the current research literature. This study addresses this gap by examining whether the content of mental health crisis text conversations that mention firearms differ from those that do not mention firearms in a large, unique dataset from a national crisis text line. MethodsWe examined data from over 3.2 million conversations between texters to Crisis Text Line and volunteer crisis counselors between September 2018 and July 2022. We used a study developed text classification machine learning algorithm that builds on natural language processing to identify and label whether crisis conversations mentioned firearms. We compared the frequency of psychosocial factors between conversations that mention firearms with those that did not. ResultsResults from a generalized linear mixed-effects model demonstrated that.conversations mentioning firearms more frequently were associated with suicide, racism, physical, sexual, emotional, and unspecified abuse, grief, concerns about a third party, substance use, bullying, gender and sexual identity, relationships, depression, and loneliness. Further, conversations mentioning firearms were less likely to be related to self-harm and eating/body image. ConclusionsThese results offer an initial glimpse of how firearms are mentioned in the context of acute mental health emergencies, which has been completely absent in prior literature. Our results are preliminary and help sharpen our understanding of contextual factors surrounding mental health emergencies where a firearm is mentioned.

Efficient CRNN: Towards end-to-end low resource Urdu text recognition using depthwise separable convolutions and gated recurrent units

In this study, a novel technique is proposed to recognize printed text in images for Urdu, a low-resource language with a scarcity of benchmark datasets. The proposed technique is called Efficient CRNN which uses depthwise separable convolutional and bidirectional gated recurrent unit layers, followed by connectionist temporal classification loss. The proposed technique is computationally more efficient than the existing text recognition techniques, requiring fewer parameters and computations. A multi-font printed Urdu text lines corpus is also presented, consisting of 245,000 text line images rendered using 7 different fonts. The corpus is called the MMU-Extension-22 and is used to train and evaluate existing state-of-the-art end-to-end text recognition techniques. Efficient CRNN is also evaluated using the proposed corpus. The proposed technique is first trained using a total of 196,000 text line images and then tested using 49,000 images. The Efficient CRNN technique achieved the minimum character and word error rates of 0.91% and 1.49% respectively for Urdu text line recognition under different settings, outperforming the existing computationally more complex techniques. The simple nature of the proposed technique not only makes it more efficient but also more robust for Urdu text line recognition, achieving a 2.23% reduced character error rate and a 71%11Percentage Decrease = 100 * (Baseline Value - Changed Value)/Baseline Value. decrease in character error rate as compared to the best performing existing Recurrent Neural Networks based technique. Also, the proposed technique outperforms Vision Transformer-based network achieving a 0.79% reduced character error rate accounting for a 41% decrease in error. Also, the Efficient CRNN has 49.16% reduced parameters compared to the baseline Vision Transformer technique.

Te Deum в кириллической транскрипции с подстрочным церковнославянским переводом

The Latin Psalter from the Chudov monastery, written in Cyrillic letters around the end of the 15th century, is a document of paramount importance for the study of the Latin cultural tradition in Muscovy. This unique manuscript seems to be linked to the fervent translation activity directed by the archbishop of Novgorod Gennadii (1484–1504). The wide space left between the lines of the main Latin text was without doubt supposed to be filled with the Church Slavic corresponding text. A Slavic “translation” was indeed inserted into the biblical cantica and other texts, among which the hymn Te Deum, traditionally (and wrongly) considered to be a joint composition by Saint Ambrose and Saint Augustine, which was very widespread in the Western Church.
 The article features the first diplomatic interlinear edition of the Slavic-Latin Te Deum, which is preceded by a short description and analysis of the Slavic text and its main linguistic peculiarities in comparison with another translation of the same work, made some decades later by Dmitry Gerasimov. Here all relevant differences between the two Slavic versions, both at the lexical and grammatical levels, are presented and shortly discussed. Further, some evident errors in the Slavic text of the Chudov Psalter cast some doubts on the possibility to consider it a real translation in the strict sense of the term; it seems rather to function as a lexico-grammatical interlinear gloss, to be read vertically as an auxiliary tool for a proper understanding of the Latin original.

Metropolitan Sanat Müzesi Örneğinde Türk Hat Sanatında Şeyh Hamdullah Ekolü

The art of calligraphy, also known as "hüsn-i hat" in Islamic culture, involves writing Islamic script in accordance with aesthetic principles. Executed using Arabic letters, this art form emphasizes the aesthetics and meaning of writing. The calligraphers, known as "hattat," adhere to rules to create Imagely pleasing scripts. The essence of calligraphy lies in expressing the aesthetics and essence of writing. Influenced by Islam's emphasis on writing and avoidance of imagery, calligraphy has become one of the most valued forms of Islamic art. This study aims to examine the contributions of the prominent Calligrapher Shaikh Hamdullah, a pioneer of Turkish Calligraphy, to the art through the example of a " thuluth and naskh hadith and poetry panel." Conducted through a literature review, this research delves into the Ottoman calligraphy, aklâm-ı sitte(six scripts), and the Shaikh Hamdullah school, focusing on his muraqqa at the Metropolitan Museum of Art. During the Ottoman era, aklâm-ı sitte styles like thuluth and naskh; muhaqqaq and rihani; tawqi and riq'a, evolved under Hamdullah's influence. His style of naskh calligraphy became the preferred method for transcribing the Quran across the Islamic world, giving rise to the Shaikh Hamdullah School. His contributions encompass Qur'an manuscripts, calligraphic panels, and muraqqas, held in prestigious collections worldwide. The tradition of granting, or certification, has preserved Shaikh's style, allowing his teachings to endure. The discussed artwork, a part of Shaikh's muraqqa album, features thuluth headings and naskh text lines drawn with ink and reed pen on marbled paper. The calligraphy is characterized by its harmonious thuluth and naskh styles, with minimal gold-painted floral motifs and borders. This study provides insights into the Shykh Hamdullah School and its impact on Turkish Calligraphy.

Handwriting-Based Text Line Segmentation from Malayalam Documents

Optical character recognition systems for Malayalam handwritten documents have become an open research area. A major hindrance in this research is the unavailability of a benchmark database. Therefore, a new database of 402 Malayalam handwritten document images and ground truth images of 7535 text lines is developed for the implementation of the proposed technique. This paper proposes a technique for the extraction of text lines from handwritten documents in the Malayalam language, specifically based on the handwriting of the writer. Text lines are extracted based on horizontal and vertical projection values, the size of the handwritten characters, the height of the text lines and the curved nature of the Malayalam alphabet. The proposed technique is able to overcome incorrect segmentation due to the presence of characters written with spaces above or below other characters and the overlapping of lines because of ascenders and descenders. The performance of the proposed method for text line extraction is quantitatively evaluated using the MatchScore value metric and is found to be 85.507%. The recognition accuracy, detection rate and F-measure of the proposed method are found to be 99.39%, 85.5% and 91.92%, respectively. It is experimentally verified that the proposed method outperforms some of the existing language-independent text line extraction algorithms.

Robust PDF Watermarking against Print-Scan Attack.

Portable document format (PDF) files are widely used in file transmission, exchange, and circulation because of their platform independence, small size, good browsing quality, and the ability to place hyperlinks. However, their security issues are also more thorny. It is common to distribute printed PDF files to different groups and individuals after printing. However, most PDF watermarking algorithms currently cannot resist print-scan attacks, making it difficult to apply them in leak tracing of both paper and scanned versions of PDF documents. To tackle this issue, we propose an invisible digital watermarking technology based on modifying the edge pixels of text strokes to hide information in PDFs, which achieves high robustness to print-scan attacks. Moreover, it cannot be detected by human perception systems. This method focuses on the representation of text by embedding watermarks by changing the features of the text to ensure that changes in these features can be reflected in the scanned PDF after printing. We first segment each text line into two sub-blocks, then select the row of pixels with the most black pixels, and flip the edge pixels closest to this row. This method requires the participation of original PDF documents in detection. The experimental results show that all peak signal-to-noise ratio (PSNR) values of our proposed method exceed 32 dB, which indicates satisfactory invisibility. Meanwhile, this method can extract the hidden information with 100% accuracy under the JPEG compression attack, and has high robustness against noise attacks and print-scan attacks. In the case of no attacks, the watermark can be recovered without any loss. In terms of practical applications, our method can be applied in the practical leak tracing of official paper documents after distribution.

Text line extraction strategy for palm leaf manuscripts

Text line segmentation is an important step in the historical document image analysis pipeline to supply useful information for recognition, keyword spotting and indexing. Many handcrafted-based and learning-based approaches have been developed to cope with text line extraction challenges. In this work we present a hybrid technique which combines a convolutional-based denoising task and heuristic Seam Carving framework. We propose the following changes to the original Seam Carving: (1) We applied adaptive slice to anticipate miss-extraction on a short text during medial seam computation. (2) We applied a triple smoothing to find the best local maxima of the smoothed horizontal projection profile which represents candidate medial seams. (3) We utilized five post-processing steps aimed at reconstructing a more precise medial seam. Three different palm leaf data sets: old Sundanese, old Balinese, and old Khmer, have been used to compare Convolutional Seam Carving (CSC) to several baseline methods, including the original Seam Carving. Experimental results show that the proposed method outperforms other current handcrafted-based baselines on all three palm leaf manuscript (PLM) data sets. On the old Sundanese data sets, CSC can produce a significant improvement of the performance rate compared to all other baseline approaches, and it also enhance the measurement on old Balinese and old Khmer datasets. In the ablation study, we discovered that a foreground extraction step is not only able to reduce noises and color degradation but also provide better separation of text region. Following that, an adaptive slice and triple smoothing approach contribute to solve the text length variation problem. Finally, post-processing steps were effective in connecting discontinuous medial seam. The code has been published in https://github.com/erickpaulus/text-line-segmentation.

Foreseeing the coming of a human being

The prose by the contemporary Russian writer Kirill Ryabov is deceptively simple and bleak. Known for their pithy and cinematographic phrases, dynamic plot and succinct portrayals of the characters’ appearance and speech,Ryabov’s books remain highly seductive for critics and offer numerous angles of interpretation. A. Gusamova’s article is written in a manner that is unmistakably Ryabov’s, i. e., only a few lines of text suffice to reveal the main message: the paradox of Ryabov’s books is that most of them remain unwritten. A two-hundred-page paperback constitutes the bare essentials.Everything omittable has been left out: descriptions, the characters’ personal traits, as well as the usual psychological particulars. Ryabov’s poetics is founded in omitted elements. Gusamova seeks to reconstruct them in her analysis and interpretation of the milestones marking Ryabov’s artistic evolution. The article focuses on the development of the existential and moral agenda which plays a key role in the artistic world of Ryabov’s prose.

Seeger Lecture, Society for Ethnomusicology Virtual Meeting 2021

Seeger Lecture, Society for Ethnomusicology Virtual Meeting 2021

DAN: A Segmentation-Free Document Attention Network for Handwritten Document Recognition.

Unconstrained handwritten text recognition is a challenging computer vision task. It is traditionally handled by a two-step approach, combining line segmentation followed by text line recognition. For the first time, we propose an end-to-end segmentation-free architecture for the task of handwritten document recognition: the Document Attention Network. In addition to text recognition, the model is trained to label text parts using begin and end tags in an XML-like fashion. This model is made up of an FCN encoder for feature extraction and a stack of transformer decoder layers for a recurrent token-by-token prediction process. It takes whole text documents as input and sequentially outputs characters, as well as logical layout tokens. Contrary to the existing segmentation-based approaches, the model is trained without using any segmentation label. We achieve competitive results on the READ 2016 dataset at page level, as well as double-page level with a CER of 3.43% and 3.70%, respectively. We also provide results for the RIMES 2009 dataset at page level, reaching 4.54% of CER. We provide all source code and pre-trained model weights at https://github.com/FactoDeepLearning/DAN.

UTTSR: A Novel Non-Structured Text Table Recognition Model Powered by Deep Learning Technology

To prevent the compilation of documents, many table documents are formatted with non-editable and non-structured texts such as PDFs or images. Quickly recognizing the contents of tables is still a challenge due to factors such as irregular formats, uneven text quality, and complex and diverse table content. This article proposes the UTTSR table recognition model, which consists of four parts: text region detection, text line detection and recognition, and table sequence recognition. For table detection, the Cascade Faster RCNN with the ResNeXt105 network is implemented, using TPS (Thin Plate Spline) transformation and affine transformation to correct the image and to improve accuracy. For text line detection, DBNET is used with Do-Conv in FPN (Feature Pyramid Networks) to speed up training. Text lines are recognized using CRNN without the CTC module, enhancing recognition performance. Table sequence recognition is based on the transformer combined with post-processing algorithms that fuse table structure sequences and unit grid content. Experimental results show that the UTTSR model outperforms the compared methods. This upgraded model significantly improves the accuracy of the previous state-of-the-art F1 score on complex tables, reaching 97.8%.

AI-ASSISTED DIGITALISATION OF HISTORICAL DOCUMENTS

Abstract. Preserving historical archival heritage involves not only physical measures to safeguard these valuable texts but also providing for their digital preservation. However, merely digitising manuscripts and codexes is not enough. A further step is needed: the digitalisation of their content, i.e. the verbatim transcription of scanned texts. This process enables the accurate preservation of their textual content, making it easier to search for information and conduct further analyses. With the help of artificial intelligence, particularly Deep Neural Networks (DNNs), automatic handwriting recognition can be performed. In this study, we employed a Convolutional Recurrent Neural Network (CRNN), an established type of DNN, to determine the minimum amount of labelled data required to automatically transcribe five different historical datasets that vary in language and time period. The results show that a Character Error Rate (CER) lower than 10% can be achieved with just a few hundred labelled text lines in almost all cases.

Deep Learning-Based Cursive Text Detection and Recognition in Natural Scene Images

Cursive text detection and recognition in natural scene images are complex tasks due to the variability and intricacy of handwriting styles. This research focuses on developing a deep learning-based approach to address these challenges. The proposed solution leverages advancements in deep learning techniques to improve the accuracy and robustness of cursive text detection and recognition. The research involves collecting and annotating a diverse dataset of natural scene images containing cursive text. A deep learning model for cursive text detection is trained on the annotated dataset, and techniques for text line segmentation are investigated to enhance accuracy. Furthermore, a deep learning-based recognition system is designed and implemented to transcribe cursive text into machine-readable text. The proposed approach is evaluated and compared with existing methods using appropriate metrics and benchmark datasets. The research aims to provide insights into the challenges and opportunities of cursive text analysis in real-world scenarios and contribute to advancements in document digitization, handwriting analysis, and information retrieval.

Point-of-Care Feedback: Preceptor Cosigning Workflow Promotes Milestone Assessment

In residency training, the Accreditation Council for Graduate Medical Education stipulates feedback from faculty should be frequent, exist in the context of routine clinical care, include multiple evaluators, emphasize competency and milestones, and be in writing so it can be shared with the Clinical Competency Committee.Collecting frequent feedback in the context of ambulatory care is an ambitious ask. Supervision of residents in clinic can be busy, if not chaotic. Assessment surveys can be quick, and they can be web-based or smartphone application-based. Still, collection rates can be lower than is needed. Faculty reminders by email, signs in the supervision area, available program links on workplace computers, and prompts submitted by residents to supervising faculty have all been used to increase assessment completion by faculty.1Oregon Health & Science University's Internal Medicine Clinic is the resident continuity clinic for 44 trainees, supervised by more than 30 part-time faculty preceptors. We tested a method to increase case-specific direct observation feedback by preceptors.We looked to thought leaders in informatics for help in our tool design. The most frequently used clinical decision support tools are directly integrated with electronic medical record (EMR) systems. To design effective point-of-care support tools, our experts stick with these informatics rules of thumb: “Speed is everything,” “Deliver in real-time,” and “Fit into the user's workflow.”2We designed a web-based rapid assessment tool. The evaluator selects content areas or processes to score and enters a narrative comment. Our innovation was to deliver this tool directly within the preceptor workflow: we embedded a link in the cosigning (preceptor attestation) template. The location of the survey link was a game changer for our collection rate.Embedding the link into the cosigning template took some care. So that our survey hyperlink would never become part of the patient chart, we needed disappearing instructional text with automated removal from the chart when the preceptor note was signed. Many EMRs enable “prompts” in templates or text blocks. Our university utilizes EpicCare, and we were able to achieve disappearing text using the function “blank, optional SmartList.” Our function displays as a single line of text at the bottom of the preceptor note in progress, with blue font that is recognized by the clinician as disappearing text. The text contains a hyperlink to our rapid assessment.Within the attestation workflow, a preceptor has a visual reminder to provide the assessment and has the option to enter the assessment survey with a single click. The survey results can be collected at any interval by the clinic medical director, who has access to survey reports.Links to our rapid assessment tool were provided in several lower yield environments before we identified a way to embed a link into the cosigning workflow (Figure). Embedding the link within the existing preceptor workflow of chart attestation resulted in an immediate, meaningful, and sustained increase in the number of submitted assessments.The feedback entered is exceptionally relevant—it is specific and contains colorful clinical performance anecdotes that inform resident competencies. This process has been popular with residents and faculty, and it has markedly enriched our resident feedback process.

Parental Rights Laws and the Impact on Mental Health Services and Counseling for Adolescents.

Parental Rights Laws and the Impact on Mental Health Services and Counseling for Adolescents.

End-to-End page-Level assessment of handwritten text recognition

The evaluation of Handwritten Text Recognition (HTR) systems has traditionally used metrics based on the edit distance between HTR and ground truth (GT) transcripts, at both the character and word levels. This is very adequate when the experimental protocol assumes that both GT and HTR text lines are the same, which allows edit distances to be independently computed to each given line. Driven by recent advances in pattern recognition, HTR systems increasingly face the end-to-end page-level transcription of a document, where the precision of locating the different text lines and their corresponding reading order (RO) play a key role. In such a case, the standard metrics do not take into account the inconsistencies that might appear. In this paper, the problem of evaluating HTR systems at the page level is introduced in detail. We analyse the convenience of using a two-fold evaluation, where the transcription accuracy and the RO goodness are considered separately. Different alternatives are proposed, analysed and empirically compared both through partially simulated and through real, full end-to-end experiments. Results support the validity of the proposed two-fold evaluation approach. An important conclusion is that such an evaluation can be adequately achieved by just two simple and well-known metrics: the Word Error Rate (WER), that takes transcription sequentiality into account, and the here re-formulated Bag of Words Word Error Rate (bWER), that ignores order. While the latter directly and very accurately assess intrinsic word recognition errors, the difference between both metrics (ΔWER) gracefully correlates with the Normalised Spearman’s Foot Rule Distance (NSFD), a metric which explicitly measures RO errors associated with layout analysis flaws. To arrive to these conclusions, we have introduced another metric called Hungarian Word Word Rate (hWER), based on a here proposed regularised version of the Hungarian Algorithm. This metric is shown to be always almost identical to bWER and both bWER and hWER are also almost identical to WER whenever HTR transcripts and GT references are guarantee to be in the same RO.

Historical document image analysis using controlled data for pre-training

Using neural networks for semantic labeling has become a dominant technique for layout analysis of historical document images. However, to train or fine-tune appropriate models, large labeled datasets are needed. This paper addresses the case when only limited labeled data are available and promotes a novel approach using so-called controlled data to pre-train the networks. Two different strategies are proposed: The first addresses the real labeling task by using artificial data; the second uses real data to pre-train the networks with a pretext task. To assess these strategies, a large set of experiments has been carried out on a text line detection and classification task using different variants of U-Net. The observations, obtained from two different datasets, show that globally the approach reduces the training time while offering similar or better performance. Furthermore, the effect is bigger on lightweight network architectures.

Nautilus

When a digital collection has been processed by OCR, the usability expectationsof patrons and researchers are high. While the former expect full textsearch to return all instances of terms in historical collections correctly, the latterare more familiar with the impacts of OCR errors but would still like toapply big data analysis or machine-learning methods. All of these use casesdepend on high quality textual transcriptions of the scans. This is why theNational Library of Luxembourg (BnL) has developed a pipeline to improveOCR for existing digitised documents. Enhancing OCR in a digital library notonly demands improved machine learning models, but also requires a coherentreprocessing strategy in order to apply them efficiently in production systems. The newly developed software tool, Nautilus, fulfils these requirements usingMETS/ALTO as a pivot format. The BnL has open-sourced it so that otherlibraries can re-use it on their own collections. This paper covers the creationof the ground truth, the details of the reprocessing pipeline, its production useon the entirety of the BnL collection, along with the estimated results. Basedon a quality prediction measure, developed during the project, approximately28 million additional text lines now exceed the quality threshold.