Unmasking the Unknown: Facial Deepfake Detection in the Open-Set Paradigm

View on arXiv ← Back to list

Authors: Nadarasar Bahavan, Sanjay Saha, Ken Chen, Sachith Seneviratne, Sanka Rasnayaka, Saman Halgamuge

Published: 2025-03-11 05:23:07+00:00

AI Summary

This paper proposes a novel open-set deepfake detection algorithm using supervised contrastive learning. Unlike closed-set methods, this approach can identify deepfakes generated by unknown techniques and flag them as 'unknown', improving robustness and reliability.

Abstract

Facial forgery methods such as deepfakes can be misused for identity manipulation and spreading misinformation. They have evolved alongside advancements in generative AI, leading to new and more sophisticated forgery techniques that diverge from existing 'known' methods. Conventional deepfake detection methods use the closedset paradigm, thus limiting their applicability to detecting forgeries created using methods that are not part of the training dataset. In this paper, we propose a shift from the closed-set paradigm for deepfake detection. In the open-set paradigm, models are designed not only to identify images created by known facial forgery methods but also to identify and flag those produced by previously unknown methods as 'unknown' and not as unforged/real/unmanipulated. In this paper, we propose an open-set deepfake classification algorithm based on supervised contrastive learning. The open-set paradigm used in our model allows it to function as a more robust tool capable of handling emerging and unseen deepfake techniques, enhancing reliability and confidence, and complementing forensic analysis. In open-set paradigm, we identify three groups including the unknown group that is neither considered known deepfake nor real. We investigate deepfake open-set classification across three scenarios, classifying deepfakes from unknown methods not as real, distinguishing real images from deepfakes, and classifying deepfakes from known methods, using the FaceForensics++ dataset as a benchmark. Our method achieves state of the art results in the first two tasks and competitive results in the third task.

Key findings

The proposed method achieves state-of-the-art results in detecting unknown deepfakes and distinguishing real images from deepfakes, and competitive results in detecting known deepfakes. The use of supervised contrastive learning proves effective in learning discriminative features for open-set classification.

Approach

The approach uses a three-stage pipeline: Stage 1 employs weighted supervised contrastive learning to learn discriminative features focusing on forgery characteristics. Stage 2 fine-tunes a classifier on top of the learned encoder. Stage 3 uses softmax thresholding on the training data to identify unknown deepfakes.

Datasets

FaceForensics++ (FF++c23), CelebDFv2

Model(s)

DenseNet121 (encoder), a linear layer (projection network), and a classifier network. Xception is used as a baseline.

Author countries

Sri Lanka, Singapore, China, Australia

← Previous