PHTI: Pashto Handwritten Text Imagebase for Deep Learning Applications

Abstract

Document Image Analysis (DIA) is one of the research areas of Artificial Intelligence (AI) that converts document images into machine-readable codes. In DIA systems, Optical Character Recognition (OCR) plays a key role in digitizing document images. The output of an OCR system is further used in many applications including, Natural Language Processing (NLP), Sentiment Analysis, Speech Recognition, and Translation Services. However, standard datasets are an essential requirement for the development, evaluation and comparison of different text recognition techniques. Pashto is one of such low resource languages that lacks availability regarding standard dataset of handwritten text. This paper therefore, addresses the unavailability of standard dataset for the Pashto handwritten text by developing a dataset named Pashto Handwritten Text Imagebase (PHTI). The PHTI is created by collecting handwritten samples from diverse genre of the Pashto language including poetry, religion, short stories, articles, novels, sports, culture and news. The dataset consists of 4, 000 scanned images, written by 400 writers including 200 males and 200 females. These 4, 000 images are further segmented into 36, 082 text-line images. Each text-line image is annotated/ transcribed with UTF-8 codecs. The dataset can be used for many deep learning-based applications including, text recognition, skew detection, gender classification and age-groups classification.

Deep learning-based recognition system for pashto handwritten text: benchmark on PHTI

“… Hussain et al (2022) published a dataset named Pashto Handwritten Text Imagebase (PHTI). The PHTI contains text line images presenting real world Pashto text in handwritten form.…”

“…This article recognizes Pashto handwritten text using MD-LSTM ( Graves & Schmidhuber, 2008 ) architecture. Several experiments figure out the best hidden layer size for the proposed MD-LSTM system as a benchmark on the PHTI dataset ( Hussain et al, 2022 ). The proposed architecture is different from the model of Graves & Schmidhuber (2008) and from the model of Ahmad et al (2016) in terms of number of hidden layer size and Tanh layer size for LSTM units.…”

Pashto handwritten character recognition based on Generative Adversarial Network (GAN) and Deep learning

“…The evaluation results show that the LSTM performs significantly better than the HMM and traditional feature extraction techniques such as SIFT, particularly in the presence of scale and rotation variations. [61]This research introduces a new dataset called KPTI for the recognition of Pashto text. It contains 17,015 images of text lines with corresponding ground truth.…”

Using the AraBERT Model for Customer Satisfaction Classification of Telecom Sectors in Saudi Arabia

“…Convolutional Neural Network (CNN), which was initially developed in the Neural Network Image Processing Community (NNIPC), is a famous kind of Feed-forward Neural Network (FNN) with a profound structure and has shown outstanding simulation results in various tasks, particularly in NLP tasks, such as sentence analysis of various languages in different applications. Multiple types of CNN, including the typical model, have been an important focus of research as they can be applied to complex problems involving time-varying patterns [49][50][51]. The standard CNN involves two operations, which feature extractors, convolution and pooling; the obtained output is then associated with the following [23].…”