<![CDATA[The increasing use of digital payments potentially elevates the risk of personal data theft and unauthorized access to applications. To mitigate this, biometric-based authentication, such as facial recognition, can be implemented. This study aims to utilize facial recognition as user authentication within an application. The facial recognition is developed using MobileNetV2. This research encompasses data collection, pre-processing, data splitting, data augmentation, model training, model evaluation, and application development. The total facial image data collected was 100 images from 5 classes with an image size of 160 x 160 pixels in .jpg format, sourced from direct photography using a smartphone camera (12MP resolution) under controlled indoor lighting conditions with consistent distance of approximately 50 cm from the subject. The model was successfully implemented with an accuracy of 85%. The model achieved successful implementation with 85% accuracy for real-time facial authentication in digital payment applications.]]>
Copyrights © 2026
