Mendeley  

Zotero  

Recognition of Afghanistan's General University Entrance Exam ID from the Exam Sheet. JOTS 2024; 2 (3) :3-14
URL: http://jots.knu.edu.af/article-1-43-en.html

The use and promotion of artificial intelligence (AI) in advancing industries is one of humanity's remarkable achievements in recent decades, applied in various fields. Today, governmental and non-governmental organizations face an urgent need for administrative automation and problem-solving through AI. Accordingly, this research aims to address the problem of identifying candidates' IDs for university entrance exams using AI.
In this study, the identification and verification of candidates' IDs from their exam sheets are carried out in two stages. In the first stage, the ID is recognized based on the digits handwritten by the candidates in the designated ID section of the exam sheet. In the second stage, the boldened keys marked in the ID section are utilized. Both stages are performed after scanning the exam sheets, employing image processing techniques and cropping operations to isolate the required areas.
For recognizing handwritten digits, the MNIST dataset and the TensorFlow and Keras libraries were used to develop a high-accuracy model. In the second stage, the key section is divided into nine columns, each containing a single key, and then rows are processed with precision. The boldened keys are extracted using a specific threshold value, and ultimately, the candidate's ID number is identified through both methods to ensure higher reliability.