Akbulut, UlasOniz, Yesim2026-04-042026-04-042025979-833159753-5https://doi.org/10.1109/ISMSIT67332.2025.11268171https://hdl.handle.net/11411/102459th International Symposium on Multidisciplinary Studies and Innovative Technologies, ISMSIT 2025 -- 14 November 2025 through 16 November 2025 -- Ankara -- 217734Machine learning techniques are widely used to recognize handwritten words or digits in different applications including check depositing and mobile banking. The main goal of this study is to employ deep learning techniques on a website to make an application that recognizes drawn Arabic digits on a canvas. The performance of the proposed Convolutional Neural Network(CNN) has been compared to the performance of a Multi Layer Perceptron Model(MLP) indicating a better recognition accuracy. To assess the performance of the proposed network two different datasets have been utilized; The widely used MNIST dataset that includes 60000 Arabic Digit images and a custom dataset collected using the developed website in Visual studio. The hyperparameters have been determined empirically. The trained network have been embedded to the generated website with an accuracy of 93.33 percent. © 2025 IEEE.eninfo:eu-repo/semantics/closedAccessConvolutional Neural Network(Cnn)Deep LearningHandwritten Digit RecognitionMulti Layer Perceptron(Mlp)Conference Paper2-s2.0-10503114849210.1109/ISMSIT67332.2025.11268171N/A