Squeeze-and-Excitation Integrated MobileNetV2 for Efficient DeepFake Detection

The rapid advancement of deepfake technology has raised significant concerns regarding the authenticity of visual content, creating an urgent need for efficient and accurate detection methods. This paper presents an enhanced MobileNetV2 architecture integrated with a Squeeze-and-Excitation (SE) block, designed to improve performance on image classification tasks, particularly for distinguishing real and fake faces. The proposed SE-enhanced MobileNetV2 model is evaluated using a dataset of 1,40,000 real and fake face images, achieving an impressive 97% accuracy on the test dataset. Comparative analyses demonstrate that the SE-enhanced MobileNetV2 outperforms baseline models, including standard MobileNetV2, EfficientNet-B0, and EfficientNet-B7, in terms of both classification accuracy and resource consumption.