Handwriting Analysis AI-based System for Assisting People with Dysgraphia - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

Handwriting Analysis AI-based System for Assisting People with Dysgraphia

Résumé

Dysgraphia is a learning disability of written expression, which affects the ability to write, mainly handwriting and coherence. Several studies have proposed approaches for assisting dysgraphic people based on AI algorithms. However, existing aids for dysgraphia take only one aspect of the problem faced by the patients into consideration. Indeed, while some provide writing assistance, others address spelling or grammatical problems. In this paper, a novel system for helping people suffering from dysgraphia is proposed. Our system tackles several problems, such as spelling mistakes, grammatical mistakes and poor handwriting quality. Further, a text-to-speech functionality is added to improve the results. The proposed system combines a plethora of solutions into a valuable approach for efficient handwriting correction: handwritten text recognition using a CNN-RNN-CTC model, a spelling correction model based on the SymSpell and Phoneme models, and a grammar correction using the GECToR model. Three machine learning models are proposed. The experimental results are compared based on the values of Character error rate, Word error rate, and the workflow of three handwritten text recognition models, and has led to an improvement of the quality of the results.
Fichier principal
Vignette du fichier
ICCS2023.pdf (998.58 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04341888 , version 1 (13-12-2023)

Identifiants

Citer

Richa Gupta, Deepti Mehrotra, Redouane Bouhamoum, Maroua Masmoudi, Hajer Baazaoui. Handwriting Analysis AI-based System for Assisting People with Dysgraphia. 23rd International Conference Computational Science – ICCS 2023, Jun 2023, Prague, France. pp.185-199, ⟨10.1007/978-3-031-36024-4_14⟩. ⟨hal-04341888⟩
27 Consultations
66 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More