<![CDATA[The rapid development of Generative AI technology enables the creation of photorealistic synthetic images and deepfake manipulations based on Generative Adversarial Networks (GAN/StyleGAN). Human cognitive visual limitations in recognizing micro-artifacts with the naked eye underscore the urgent need for an automated computing system capable of verifying image authenticity objectively, transparently, and practically. This study aims to design and integrate an Artificial Intelligence model based on a Convolutional Neural Network (CNN) with MobileNetV2 Transfer Learning architecture into a web application interface powered by the Laravel 12 framework. The software design lifecycle was managed using the Adaptive Scrum Model methodology. The results demonstrated that training the static image detection model using a 140k faces dataset successfully locked optimal weights at Epoch 12, achieving a validation accuracy of 85.57%. Integration via Explainable AI (Grad-CAM) effectively visualized pixel anomaly coordinates. Functional testing (Black Box) proved that the system design is highly capable and ready for future deployment.]]>
Copyrights © 2026
