Reproduce “Detecting Facial Image Manipulations with Multi-Layer CNN Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Alejandro Marco Montejano, Angela Sanchez Perez, Javier Barrachina, David Ortiz-Perez, Manuel Benavent-Lledo, Jose Garcia-Rodriguez
Researchers from Facephi and the University of Alicante developed enhanced multi-layer CNNs for detecting and classifying facial image manipulations, achieving up to 76% overall accuracy in distinguishing genuine, DeepFake, and FaceSwap images, including those generated by Stable Diffusion. The models maintain computational efficiency while improving generalization over previous compact networks.
Key insights
- Increasing the depth of compact CNN architectures significantly improves both generalization to unseen data and performance in multi-class facial manipulation detection.
- Training detection models on diverse datasets, including images generated by advanced AI models like Stable Diffusion, is essential for robustness against evolving manipulation techniques.
- Lightweight CNN models can be effectively adapted for multi-class facial manipulation detection, offering computationally efficient solutions suitable for practical deployment.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.