Advanced Deep Learning Techniques for Automated Fake Image Analysis and Classification

The rapid advancement of Artificial Intelligence (AI) has enabled the generation of highly realistic synthetic images, creating significant challenges in digital media authentication, cyber security, journalism, and social networking platforms. Existing fake image detection methods often struggle to distinguish AI-generated images from authentic photographs due to the increasing quality of generative models. This work proposes an automated fake image classification framework using deep learning techniques. Three different deep learning architectures—Convolutional Neural Network (CNN), MobileNetV2, and EfficientNetB0—were implemented and evaluated using the CIFAKE dataset enriched with additional real-world images collected from mobile cameras, WhatsApp, screenshots, and internet sources. Experimental results demonstrate that EfficientNetB0 significantly outperforms CNN and MobileNetV2 by achieving an accuracy of 95.54%, precision of 97.27%, recall of 93.70%, and F1-score of 95.45%. The proposed framework provides an effective and computationally efficient solution for automated fake image detection. Keywords— Deep Learning, Fake Image Detection, EfficientNetB0, MobileNetV2, CNN, Artificial Intelligence, Image Classification.