Revisiting the Efficacy of Image-Based Recognition of Arabic Handwriting: The Case of Isolated Letters

This paper investigates the performance of Machine Learning techniques for handwritten Arabic character recognition. The study employs two methods using CNNs (Convolutional Neural Networks) in one model, and a combination of CNNs and RNNs (Recurrent Neural Networks) in a Hybrid model to recognize handwritten Arabic characters from the AHCD dataset. The paper presents a comparative analysis of the proposed models, exploring their strengths and limitations. The experimental results demonstrate that the hybrid model is more accurate in recognizing handwritten Arabic characters, achieving an average accuracy rate of 91.90%.