Arabic Handwritten Documents Segmentation into Text-Lines and Words using Deep Learning

Neche, Chemseddine; Belaïd, Abdel; Kacem-Echi, Afef

doi:10.1109/icdarw.2019.50110

Cited by 29 publications

(17 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With deep learning approaches, semantic segmentation is widely used. In (Neche et al, 2019) the authors propose to segmentation of a historical document into text line by using the Residual U-Net architecture and classify pixels of the input image into three classes: background, paragraphs, and lines of text. In (Aïcha Gader and Echi, 2020) also use the modified U-Net architecture by integrating a recurrent residual convolutional neural network (RRCNN) with an attention mechanism called AR2U-Net to find precise features in a specific region, and the performance on BADAM dataset gives 93.7 % of F-measure.…”

Section: Text Line Extractionmentioning

confidence: 99%

“…The authors in (Neche et al, 2019) use deep learning for word segmentation by putting the text line image into a CNN, which extracts the features that sequentially pass to the bidirectional long short-term memory (BLSTM) network followed by CTC function (Graves et al, 2006) in order to find the alignment sequence between words and spaces. The alignment result on the lines of KHATT dataset reaches 80.1% of the F-measure.…”

Section: Word Extractionmentioning

confidence: 99%

See 1 more Smart Citation

Text line and word detection and recognition of historical Arabic manuscripts

Hakim¹,

Belaid²

2023

Preprint

View full text Add to dashboard Cite

The old Arabic manuscripts are highly sought-after documents but very difficult to access. Digitization, and thus handwriting recognition, is a beneficial way to make these resources accessible. This paper presents an end-to-end approach to the offline recognition of ancient manuscripts. First, a crucial pre-processing step is to extract text lines and words by applying transfer learning on YOLO (You Only Look Once) architecture. Thus the segmentation problem is treated as a detection problem. Then for the recognition of old handwritten words, we propose ensemble learning techniques based on recurrent neural networks associated with the Connectionist Temporal Classification layer (CTC) combined to convolution networks with Squeeze-and-Excitation blocks. The presented work accurately detects lines of text and words, even when overlapping or touching words are present, and correctly identifies those with multiple connected components. We evaluate this approach on a collection of 20 pages for text line detection. Moreover, we introduce a new consistent and accurate dataset for word detection and recognition. We have achieved promising results with 98.1% and 94.38% F1-measure on the text line and word detection, respectively, with a character error rate recognition of 8.27%.

show abstract

Section: Text Line Extractionmentioning

confidence: 99%

Section: Word Extractionmentioning

confidence: 99%

Text line and word detection and recognition of historical Arabic manuscripts

Hakim¹,

Belaid²

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The detection of text lines has been widely explored in historical manuscript text books [26,9] and other historical documents of different natures, such as newspapers [25], meteorological tables [1] finding aids [33], as well as many other supports. With index tables, one can consider the issue as a two-class image segmentation task: we separate text lines from the background.…”

Section: Document Image Analysismentioning

confidence: 99%

Text Line Detection in Historical Index Tables: Evaluations on a New French PArish REcord Survey Dataset (PARES)

Bernard,

Wall,

Boillet

et al. 2023

Lecture Notes in Computer Science

View full text Add to dashboard Cite

In this paper, we address the challenge of document image analysis for historical index table documents with handwritten records. Demographic studies can gain insight from the use of automatic document analysis in such documents through the study of population movements. To evaluate the efficacy of automatic layout analysis tools, we release the PARES dataset [6], which contains 250 labeled index table images originating from French archives. Also, we run state-of-the-art algorithms (U-FCN, R-CNN and Transformers) in order to detect the lines within index tables, a common prerequisite for handwritten text recognition (HTR). Our results indicate that text line extraction works well with the U-FCN model, while also indicating that Transformer architectures show promise for accurate text line detection in such historical documents with great efficiency. This is a encouraging step towards a Transformer-based architecture for both layout and content detection. This process and dataset represent a first step to automatically analyze handwritten and historical index tables. In addition to this paper and the PARES [6] dataset of historical index tables of 250 images, we release segmentation masks, the code we used to train and test the models, and the models themselves.

show abstract

“…The baseline definition was modified slightly towards manuscripts written in Arabic scripts. Mechi et al [28] and Neche et al [29] used an U-net and RU-net deep-learning models, which are variants of FCN. The models are trained for X-height based pixel-wise classifications of text lines.…”

Section: Related Workmentioning

confidence: 99%

Unsupervised deep learning for text line segmentation

Barakat

Droby

Alaasam

et al. 2021

2020 25th International Conference on Pattern Recognition (ICPR)

View full text Add to dashboard Cite

We present an unsupervised text line segmentation method that is inspired by the relative variance between text lines and spaces among text lines. Handwritten text line segmentation is important for the efficiency of further processing. A common method is to train a deep learning network for embedding the document image into an image of blob lines which are tracing the text lines. Previous methods learned such embedding in a supervised manner, requiring the annotation of many document images. This paper presents an unsupervised embedding of document image patches without a need of annotations. The main idea is that the number of foreground pixels over the text lines is relatively different from the number of foreground pixels over the spaces among text lines. Generating similar and different pairs relying on this principle definitely leads to outliers. However, as the results show, the outliers do not harm the convergence and the network learns to discriminate the text lines from the spaces between text lines. We experimented with a challenging Arabic handwritten text line segmentation dataset, VML-AHTE, and achieved a superior performance even over the supervised methods.

show abstract

Arabic Handwritten Documents Segmentation into Text-Lines and Words using Deep Learning

Cited by 29 publications

References 24 publications

Text line and word detection and recognition of historical Arabic manuscripts

Text line and word detection and recognition of historical Arabic manuscripts

Text Line Detection in Historical Index Tables: Evaluations on a New French PArish REcord Survey Dataset (PARES)

Unsupervised deep learning for text line segmentation

Contact Info

Product

Resources

About