Penny-Wise and Pound-Foolish in Deepfake Detection

TL;DR

This paper introduces PoundNet, a novel framework that enhances deepfake detection generalization by balancing knowledge retention and task-specific learning, significantly outperforming existing methods across multiple datasets.

Contribution

PoundNet employs a learnable prompt and balanced objective to improve deepfake detection generalization while preserving upstream task knowledge, addressing limitations of prior specialized fine-tuning approaches.

Findings

Achieves 19% improvement in deepfake detection accuracy over state-of-the-art methods.

Maintains 63% performance on object classification tasks, demonstrating knowledge retention.

Evaluated across 10 large-scale datasets, forming the largest benchmark for generalization assessment.

Abstract

The diffusion of deepfake technologies has sparked serious concerns about its potential misuse across various domains, prompting the urgent need for robust detection methods. Despite advancement, many current approaches prioritize short-term gains at expense of long-term effectiveness. This paper critiques the overly specialized approach of fine-tuning pre-trained models solely with a penny-wise objective on a single deepfake dataset, while disregarding the pound-wise balance for generalization and knowledge retention. To address this "Penny-Wise and Pound-Foolish" issue, we propose a novel learning framework (PoundNet) for generalization of deepfake detection on a pre-trained vision-language model. PoundNet incorporates a learnable prompt design and a balanced objective to preserve broad knowledge from upstream tasks (object classification) while enhancing generalization for downstream tasks (deepfake detection). We train PoundNet on a standard single deepfake dataset, following common practice in the literature. We then evaluate its performance across 10 public large-scale deepfake datasets with 5 main evaluation metrics-forming the largest benchmark test set for assessing the generalization ability of deepfake detection models, to our knowledge. The comprehensive benchmark evaluation demonstrates the proposed PoundNet is significantly less "Penny-Wise and Pound-Foolish", achieving a remarkable improvement of 19% in deepfake detection performance compared to state-of-the-art methods, while maintaining a strong performance of 63% on object classification tasks, where other deepfake detection models tend to be ineffective. Code and data are open-sourced at https://github.com/iamwangyabin/PoundNet.