DEEP LEARNING-BASED AUTOMATED DETECTION OF COPYING BEHAVIOR IN EXAMINATIONS

With the widespread adoption of digital assessment platforms, the need for robust methods to detect academic dishonesty, such as copying during exams, has become increasingly crucial. This research presents a novel approach utilizing Convolutional Neural Networks (CNNs) for the automated classification of student copying behaviors in examination scenarios. Our Convolutional Neural Network architecture is designed to capture spatial dependencies and hierarchical features within the input data, enabling it to discern subtle similarities indicative of copying. The model is trained on a diverse dataset comprising legitimate individual responses and instances of suspected copying, allowing it to learn complex patterns associated with both genuine and dishonest behavior. The proposed deep learning solution holds promise for educational institutions seeking efficient and automated methods to preserve academic integrity, providing a proactive approach to maintaining a fair and honest examination environment. As technological advancements continue to reshape education, the integration of such intelligent systems can contribute significantly to the ongoing efforts to uphold the credibility of academic assessments.

Keywords: Image Processing, examination hall Dataset, Computer vision, Convolutional Neural Network, Deep Learning.