Offline optical character recognition of Persian letters, using deep learning and feature extraction (for educational programming purposes)
This project was completed in 2010. No special libraries were used in it, so it was good practice for us to learn image processing, computer vision, and convolutional neural networks from scratch in middle school.
Read this document for more details (in Persian).
Open the project using Embarcadero RAD Studio Delphi.
Clone the project and run OCR.exe (on Windows).