Davide Coccomini

TL;DR: The Face Deepfake Detection and Reconstruction Challenge was described and two different tasks were proposed to the participants: creating a Deepfake detector capable of working in an “in the wild” scenario; and creating a method capable of reconstructing original images from Deepfakes.

TL;DR: From the authors' experiments, it emerges that EfficientNetV2 has a greater tendency to specialize often obtaining better results on training methods while Vision Transformers exhibit a superior generalization ability that makes them more competent even on images generated with new methodologies.

TL;DR: In this paper , the authors carried out an analysis of different deep learning architectures in an attempt to understand which one is more capable of better generalizing the concept of deepfake, while various deep-learning detection systems have been developed, they often struggle to detect deep-fake in real-world situations.

TL;DR: This work proposes a method useful to automatically identify misbehaving users, i.e. the users that introduce inaccuracies during the labeling phase, and detects misbehaved users as anomalies with respect to accurate ones.