Development of Deep Learning Models Using CNN Methods for Fingerprint Biometric Identification in Attendance Systems

Abstract

With the development of technology, fingerprint biometric systems are increasingly being used because they have unique and permanent patterns that are effective for identifying individuals. However, most attendance systems still rely on template matching methods, which have limitations, such as sensitivity to noise, fingerprint position variations, lack of scalability, and inability to detect fake fingerprints. As a solution, this study developed a deep learning model based on Convolutional Neural Network (CNN) to improve the accuracy and reliability of attendance systems. The methods used include fingerprint data collection and preprocessing, CNN architecture design, and model training with an adequate dataset. Evaluation was conducted using accuracy, precision, recall, and F1 score. The test results show that the designed CNN model is capable of achieving an accuracy rate of 91.68%, making the resulting attendance system more secure, efficient, and accurate than conventional methods, and can be applied on cam puses and in workplaces.