CALText: Contextual Attention Localization for Offline Handwritten Text

Anjum, Tayaba; Khan, Nazar

doi:10.1007/s11063-023-11258-5

Cited by 5 publications

(2 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similarly, for the Urdu dataset, we carefully selected both handwritten and printed publicly available datasets. The NUST-UHWR ( ul Sehr Zia et al, 2022 ) and PUCIT-OHUL ( Anjum & Khan, 2020 , 2023 ) datasets were used for handwritten text, while the UPTI dataset (which contains 0.5 million samples) and UPTI v2 dataset (which contains 1 million samples) were used for printed text. We made sure to select datasets that are diverse and representative of the Urdu language, and we took care to ensure that the data was appropriately labeled and formatted for our research purposes.…”

Section: Methodsmentioning

confidence: 99%

“…To address the scarcity of data, we combined self-collected data from 30 individuals. In addition to this, we integrated publicly accessible datasets such as NUST-UHWR ( ul Sehr Zia et al, 2022 ), PUCIT-OHUL ( Anjum & Khan, 2020 , 2023 ) and the printed Urdu vocabulary dataset. The combination of diverse datasets contributes to a more robust and versatile model.…”

Section: Proposed Approachmentioning

confidence: 99%

See 1 more Smart Citation

Adapting multilingual vision language transformers for low-resource Urdu optical character recognition (OCR)

Cheema,

Shaiq,

Mirza

et al. 2024

PeerJ Computer Science

View full text Add to dashboard Cite

In the realm of digitizing written content, the challenges posed by low-resource languages are noteworthy. These languages, often lacking in comprehensive linguistic resources, require specialized attention to develop robust systems for accurate optical character recognition (OCR). This article addresses the significance of focusing on such languages and introduces ViLanOCR, an innovative bilingual OCR system tailored for Urdu and English. Unlike existing systems, which struggle with the intricacies of low-resource languages, ViLanOCR leverages advanced multilingual transformer-based language models to achieve superior performances. The proposed approach is evaluated using the character error rate (CER) metric and achieves state-of-the-art results on the Urdu UHWR dataset, with a CER of 1.1%. The experimental results demonstrate the effectiveness of the proposed approach, surpassing state of the-art baselines in Urdu handwriting digitization.

show abstract

Section: Methodsmentioning

confidence: 99%