A Novel Hybrid DL Model for Printed Arabic Word Recognition based on GAN

The recognition of printed Arabic words remains an open area for research since is among the most complex languages. Prior has shown that few efforts have been made to develop models accurate recognition, as these faced increasing complexity performance and lack benchmark datasets. Meanwhile, Deep learning models, such Convolutional Neural Networks (CNNs), be beneficial in reducing error rate enhancing accuracy character systems. reliability increases with depth layers. Still, essential condition more layers extensive amount data. Since CNN generates features by analysing large amounts data, its directly proportional volume DL are considered data-hungry algorithms. Nevertheless, this technique suffers from poor generalisation ability overfitting issues, which affect models' accuracy. These issues due limited availability databases terms accessibility size, led a central problem facing language nowadays. Therefore, still gaps need bridged. Learning techniques also improved increase manipulating strength neural network handling datasets model building. To solve problems, study proposes hybrid word adapting deep convolutional (DCNN) work classifier based on generative adversarial (GAN) data augmentation robust improving ability. Each proposed separately evaluated compared other state-of-the-art models. tested text image dataset (APTI). shows excellent regarding accuracy, score 99.76% 94.81% DCNN APTI dataset. indicates highly competitive enhanced existing results demonstrate networks improved. This output comparable contributes body knowledge.