Browsing by Author "Supervisor: Bentrcia, Rahima"

Now showing 1 - 3 of 3
  • Thumbnail Image
    ItemOpen Access
    Arabic Handwritten Letters Recognition
    (UNIVERSITY of M'SILA, 2022-06-10) Lamri, Hamza; Laidoune, Zakaria; Supervisor: Bentrcia, Rahima
    The main goal of our work is to exploit the effectiveness of artificial intelligence methods to recognize handwritten Arabic letters. In this work, for the dataset, we use the AHCD dataset to train and test the Random Forest model. We rely on two different approaches to recognize the letter. In the first one, the model takes as inputs 16 features extracted from the image, whereas in the second approach, the model takes the whole image as an input. We use Python as the programming language. We achieved a training accuracy of 97.14% and a testing accuracy of 73.48%.
  • Thumbnail Image
    ItemOpen Access
    HANDWRITTEN DIGITS RECOGNITION
    (UNIVERSITY of M'SILA, 2022-06-10) Dilmi, Anes; Larbaoui, Ayoub Fateh Allah; Supervisor: Bentrcia, Rahima
    The main purpose of this thesis is to build a method for automatic recognition of handwritten digits. To accomplish the recognition task, first, the data set is presented as input, this is followed by a preprocessing phase, where the image undergoes various operations such as noise reduction, normalization, smoothing, and skeletonization. The result of this stage can be given as input to the stage of automatically extracting feature maps from images using 3 layers of Convolutional Neural Networks (CNN), and then fed to 4 layers of FC ANN, where these images are classified into ten classes (0 to 9). The accuracy of the final model is important because more accurate models make better decisions. Our results showed a classification accuracy of 99.78% and a loss value of 1.82% using the MNIST data set.
  • Thumbnail Image
    ItemOpen Access
    Recognition of arabic handwritten letters using deep Learning approach
    (University of M'sila, 2023-06-10) Chiekhaoui, Marouane; Aichouche, Ali; Supervisor: Bentrcia, Rahima
    Many languages have made significant advancements in the field of character recognition, including English, Chinese, Japanese, and French, with recognition rates reaching up to 100% in some cases. However, Arabic handwriting recognition faces lower recognition rates, primarily due to certain linguistic characteristics that make the recognition process more challenging, along with a shortage of high-quality available datasets. Therefore, this memorandum was undertaken with the aim of developing a system for recognizing handwritten Arabic characters and a word segmentation system. The study began with an analysis of the Arabic language's structure, followed by an overview of deep neural network technology, which has proven its efficiency in achieving rapid and reliable recognition results. Finally, the obtained results were explained and interpreted.