In this project, a dataset of handwritten Persian digits has been identified using a Multi-Layer Perceptron (MLP) neural network. In this repository, you can find the source code and results related to the training and evaluation of this neural network.
The dataset used for this exercise consists of 1,698 images of handwritten Persian digits from 0 to 9. This dataset is a subset of the comprehensive Hoda dataset (collected from the national university entrance exam in 2005 - 1384), which is provided in a file named "Hoda_Dataset.zip" along with other accompanying materials.
- Utilization of MLP for the recognition of handwritten Persian digits.
- Plotting the error curve during the training phases.
- Explanation of the maximum achieved accuracy in this project.
- Examination of accuracy in recognizing different digits.
We welcome any contributions and suggestions for improving this project. If you have ideas or enhancements, please create a new issue or send a pull request.
Note: Ensure that the Hoda_Dataset.zip file is extracted from its compressed form for the code to run correctly.
Author: Hossein Ghorbani