Hybrid Deep Learning and Machine Learning Approach for Detecting DeepFake Content

The rapid growth of artificial intelligence and deep learning technologies has led to the widespread creation of deepfake images. These manipulated images pose serious threats to digital authenticity, privacy, and the reliability of online information. Manual identification of such manipulated content has become increasingly difficult due to improvements. This paper proposes a hybrid deep learning approach for detecting deepfake images using DenseNet121, ResNet18, EfficientNet‑B4 and a custom Random CNN model. The system is trained on approximately 30,000 images from two datasets. Each model extracts visual features independently and their predictions are combined using an ensemble learning strategy. Experimental results show that the proposed hybrid model achieves an accuracy of approximately 96–98%, demonstrating strong performance in identifying manipulated images. The system is implemented using Python and Flask with a web interface that allows users to upload images and obtain prediction results.