CNN-based model on potential dyslexia detection based on automated handwriting features extraction
Children with dyslexia have additional difficulties in their academic performance and overall wellbeing. However, dyslexia is still challenging to identify quickly, slowing support and early intervention. This innovation provides a method for dyslexia identification by developing an automated system...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | en |
| Published: |
Universiti Teknologi MARA, Kedah
2023
|
| Subjects: | |
| Online Access: | https://ir.uitm.edu.my/id/eprint/128103/1/128103.pdf https://ir.uitm.edu.my/id/eprint/128103/ https://ispike.uitm.edu.my |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850998176631226368 |
|---|---|
| author | Ramlan, Siti Azura Isa, Iza Sazanita Ismail, Ahmad Puad Osman, Muhammad Khusairi Che Soh, Zainal Hisham |
| author_facet | Ramlan, Siti Azura Isa, Iza Sazanita Ismail, Ahmad Puad Osman, Muhammad Khusairi Che Soh, Zainal Hisham |
| author_sort | Ramlan, Siti Azura |
| building | Tun Abdul Razak Library |
| collection | Institutional Repository |
| content_provider | Universiti Teknologi Mara |
| content_source | UiTM Institutional Repository |
| continent | Asia |
| country | Malaysia |
| description | Children with dyslexia have additional difficulties in their academic performance and overall wellbeing. However, dyslexia is still challenging to identify quickly, slowing support and early intervention. This innovation provides a method for dyslexia identification by developing an automated system in response to this drawback. This study aims to build an automated method for potentially detecting dyslexia using automated handwriting image extraction using transfer learning Convolutional Neural Networks (CNN) by tuning the hyperparameter suited to handwriting images. The Residual network, a pretrained CNN architecture, is implemented to extract significant features automatically from handwriting images and classify them as predictive of potential dyslexia or not. The results showed impressive accuracy in classifying the handwriting images, with a testing accuracy of 90.38%. Higher accuracy percentages achieved in both the training and testing stages highlight the promise of the proposed automated dyslexia diagnosis system. Early detection of dyslexia allows for more immediate support and interventions, which leads to better educational outcomes and emotional well-being for impacted children. Furthermore, by expediting the screening process and ensuring that resources are adequately directed, the suggested automated approach can ease the load on educators and healthcare personnel. The development of an automated method for dyslexia detection based on CNN-based handwriting feature extraction represents a promising step forward in the field. The excellent accuracy rates demonstrate the proposed system's potential to improve the lives of children with dyslexia and build a more inclusive and supportive learning environment. With its high accuracy rates, the approach offers promise as an efficient tool for theearly identification of dyslexia in children, allowing for earlier intervention and targeted educational support. This research has the potential to affect both society and the education sector. |
| format | Article |
| id | my.uitm.ir-128103 |
| institution | Universiti Teknologi Mara |
| language | en |
| publishDate | 2023 |
| publisher | Universiti Teknologi MARA, Kedah |
| record_format | eprints |
| spelling | my.uitm.ir-1281032025-12-07T06:19:38Z https://ir.uitm.edu.my/id/eprint/128103/ CNN-based model on potential dyslexia detection based on automated handwriting features extraction Ramlan, Siti Azura Isa, Iza Sazanita Ismail, Ahmad Puad Osman, Muhammad Khusairi Che Soh, Zainal Hisham Educational technology Learning. Learning strategies Learning ability Children with dyslexia have additional difficulties in their academic performance and overall wellbeing. However, dyslexia is still challenging to identify quickly, slowing support and early intervention. This innovation provides a method for dyslexia identification by developing an automated system in response to this drawback. This study aims to build an automated method for potentially detecting dyslexia using automated handwriting image extraction using transfer learning Convolutional Neural Networks (CNN) by tuning the hyperparameter suited to handwriting images. The Residual network, a pretrained CNN architecture, is implemented to extract significant features automatically from handwriting images and classify them as predictive of potential dyslexia or not. The results showed impressive accuracy in classifying the handwriting images, with a testing accuracy of 90.38%. Higher accuracy percentages achieved in both the training and testing stages highlight the promise of the proposed automated dyslexia diagnosis system. Early detection of dyslexia allows for more immediate support and interventions, which leads to better educational outcomes and emotional well-being for impacted children. Furthermore, by expediting the screening process and ensuring that resources are adequately directed, the suggested automated approach can ease the load on educators and healthcare personnel. The development of an automated method for dyslexia detection based on CNN-based handwriting feature extraction represents a promising step forward in the field. The excellent accuracy rates demonstrate the proposed system's potential to improve the lives of children with dyslexia and build a more inclusive and supportive learning environment. With its high accuracy rates, the approach offers promise as an efficient tool for theearly identification of dyslexia in children, allowing for earlier intervention and targeted educational support. This research has the potential to affect both society and the education sector. Universiti Teknologi MARA, Kedah 2023 Article PeerReviewed text en https://ir.uitm.edu.my/id/eprint/128103/1/128103.pdf Ramlan, Siti Azura and Isa, Iza Sazanita and Ismail, Ahmad Puad and Osman, Muhammad Khusairi and Che Soh, Zainal Hisham (2023) CNN-based model on potential dyslexia detection based on automated handwriting features extraction. (2023) International Exhibition & Symposium On Productivity, Innovation, Knowledge & Education <https://ir.uitm.edu.my/view/publication/International_Exhibition_=26_Symposium_On_Productivity,_Innovation,_Knowledge_=26_Education.html>. pp. 584-588. ISSN 9789672948568 https://ispike.uitm.edu.my |
| spellingShingle | Educational technology Learning. Learning strategies Learning ability Ramlan, Siti Azura Isa, Iza Sazanita Ismail, Ahmad Puad Osman, Muhammad Khusairi Che Soh, Zainal Hisham CNN-based model on potential dyslexia detection based on automated handwriting features extraction |
| title | CNN-based model on potential dyslexia detection based on automated handwriting features extraction |
| title_full | CNN-based model on potential dyslexia detection based on automated handwriting features extraction |
| title_fullStr | CNN-based model on potential dyslexia detection based on automated handwriting features extraction |
| title_full_unstemmed | CNN-based model on potential dyslexia detection based on automated handwriting features extraction |
| title_short | CNN-based model on potential dyslexia detection based on automated handwriting features extraction |
| title_sort | cnn-based model on potential dyslexia detection based on automated handwriting features extraction |
| topic | Educational technology Learning. Learning strategies Learning ability |
| url | https://ir.uitm.edu.my/id/eprint/128103/1/128103.pdf https://ir.uitm.edu.my/id/eprint/128103/ https://ispike.uitm.edu.my |
| url_provider | http://ir.uitm.edu.my/ |