Integrated Ensemble UNet Architecture for Deep Fake Detection in Videos
Gadhari Pranusha, V Naveen Kumar · 2025
The proliferation of deepfake technology has raised significant concerns regarding the authenticity of digital media. Traditional deepfake detection methods often rely on single-modal approaches, which may not effectively capture the complex artifacts introduced by deepfake generation techniques. This paper proposes a novel deepfake detection framework that integrates Hybrid U-Net logic with an ensemble attention model, leveraging both spatial and temporal features for enhanced detection accuracy. The proposed system employs a hybrid architecture combining convolutional neural networks (CNNs) with U-Net structures and attention mechanisms, facilitating the extraction of both local and global features from video frames. Ensemble learning further improves robustness and generalization across diverse datasets.