Recognizing Handwritten Arabic Characters Using Deep Learning Techniques in Educational Platforms | ||
| المجلة العلمية لكلية التربية النوعية جامعة دمياط | ||
| Volume 2024, Issue 9, June 2024, Pages 324-344 PDF (1.85 M) | ||
| Document Type: البحوث العلمية الأصيلة | ||
| DOI: 10.21608/sjeud.2024.284278.1032 | ||
| Authors | ||
| Alaa Abood* 1; Elsaeed AbdElrazek2; Abeer M Saad3 | ||
| 1Damietta University. Faculty of Specific Education. Computer Department | ||
| 2Damietta University, Faculty of Specific Education, , Computer Department | ||
| 3Damietta university Faculty of Specific Education, , Computer Department | ||
| Abstract | ||
| Handwriting is a fascinating aspect of human communication that embodies the complex interaction of cognitive and motor abilities and cultural expression. Handwriting has been an enduring tribute to human inventiveness and ability, dating back to the first writings on cave walls and the exquisite calligraphy of medieval manuscripts. Handwriting, in addition to its aesthetic appeal and historical value, displays God's immense power in His creation, as seen by the exquisite design of the human hand and brain. The researchers focus their work on recognizing several linguistic handwritings. One of the languages that is still difficult for the researcher to recognize is Arabic handwriting because of a number of its characteristics, such as connectivity, the presence of dots, and diacritical marks. This research presents the development of a system based on the recognition of Arabic handwritten characters in education platforms using three deep learning-based models. The suggested models, a pre-trained CNN (VGG-16, MobileNet) and a convolutional neural network (CNN), were trained on the AHCD dataset, which was created by 60 authors ranging in age from 19 to 40. The experiment's findings demonstrated CNN was better than the others, with 96.4% accuracy on the test set, compared to 95% accuracy for MobileNet and 90% accuracy for VGG16. | ||
| Keywords | ||
| Handwritten character recognition; deep learning; convolutional neural network; VGG-16; MobileNet | ||
|
Statistics Article View: 137 PDF Download: 336 |
||
