Summary of Deepfeaturex Net: Deep Features Extractors Based Network For Discriminating Synthetic From Real Images, by Orazio Pontorno (1) et al.
DeepFeatureX Net: Deep Features eXtractors based Network for discriminating synthetic from real images
by Orazio Pontorno, Luca Guarnera, Sebastiano Battiato
First submitted to arxiv on: 24 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed approach to deepfake detection is a novel three-block architecture that utilizes deliberately unbalanced datasets to train Base Models for extracting discriminative features from real, GAN-generated, and Diffusion Model-generated images. The concatenated features are then processed to determine the origin of an input image. Experimental results show improved robustness to JPEG compression and better generalization performance compared to state-of-the-art methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Deepfakes are fake images created by deep learning algorithms that can be very difficult to detect. Scientists are working on ways to figure out if an image is real or fake. The problem is that these methods often don’t work well when they’re tested with images that were made using a different algorithm than the one used to train them. This makes it hard for the method to do well in real-world situations. In this paper, researchers propose a new way to detect deepfakes by dividing the task into three parts and training each part separately using special datasets that are designed to help the models learn what features to look for. They test their approach and find that it does better than other methods at detecting deepfakes, even when the images have been compressed. |
Keywords
» Artificial intelligence » Deep learning » Diffusion model » Gan » Generalization
