[论文解读] FDFtNet: Facing Off Fake Images using Fake Detection Fine-tuning Network
FDFtNet 是一种轻量级、微调型神经网络架构,通过仅使用少量真实与虚假图像,显著提升预训练CNN在鲁棒虚假图像检测中的性能。通过集成一种新型基于自注意力机制的微调Transformer(FTT)模块和MobileNet V3模块,其在Deepfake数据集上达到97.02%的最先进准确率,在GAN生成数据集上达到90.29%,显著优于基线模型,且所需数据极少。
Creating fake images and videos such as "Deepfake" has become much easier these days due to the advancement in Generative Adversarial Networks (GANs). Moreover, recent research such as the few-shot learning can create highly realistic personalized fake images with only a few images. Therefore, the threat of Deepfake to be used for a variety of malicious intents such as propagating fake images and videos becomes prevalent. And detecting these machine-generated fake images has been quite challenging than ever. In this work, we propose a light-weight robust fine-tuning neural network-based classifier architecture called Fake Detection Fine-tuning Network (FDFtNet), which is capable of detecting many of the new fake face image generation models, and can be easily combined with existing image classification networks and finetuned on a few datasets. In contrast to many existing methods, our approach aims to reuse popular pre-trained models with only a few images for fine-tuning to effectively detect fake images. The core of our approach is to introduce an image-based self-attention module called Fine-Tune Transformer that uses only the attention module and the down-sampling layer. This module is added to the pre-trained model and fine-tuned on a few data to search for new sets of feature space to detect fake images. We experiment with our FDFtNet on the GANsbased dataset (Progressive Growing GAN) and Deepfake-based dataset (Deepfake and Face2Face) with a small input image resolution of 64x64 that complicates detection. Our FDFtNet achieves an overall accuracy of 90.29% in detecting fake images generated from the GANs-based dataset, outperforming the state-of-the-art.
研究动机与目标
- 为应对由GAN和少样本学习技术生成的高质量虚假图像日益增长的威胁。
- 开发一种轻量级、可泛化的虚假图像检测方法,仅需少量训练样本。
- 在不从零开始训练的前提下,提升现有预训练CNN在虚假图像检测中的性能。
- 通过高效微调实现对新型虚假图像生成模型的快速适应。
提出的方法
- 提出一种新型基于图像的自注意力模块——微调Transformer(FTT),仅使用注意力机制和下采样层来学习新的特征表示。
- 在移除原始分类头后,将FTT模块与预训练CNN主干网络(如Xception、ResNetV2、ShallowNetV3)集成。
- 采用MobileNet V3模块(MBblockV3)通过深度可分离卷积和高效架构增强特征提取。
- 通过Cutout数据增强方法(超参数α=3,β=10)提升小样本数据集上的鲁棒性。
- 在有限的真实与虚假图像对上端到端微调整个FDFtNet架构,使用标准交叉熵损失。
- 采用全局平均池化(GAP)后接最终分类层,预测图像为真实或虚假的标签。
实验结果
研究问题
- RQ1一种轻量级、基于微调的神经网络架构能否在极少训练数据下有效检测由现代GAN和少样本方法生成的多样化虚假图像?
- RQ2与标准预训练模型微调相比,集成自注意力模块(FTT)在检测性能上有多大提升?
- RQ3FDFtNet在不同虚假图像生成技术(如Progressive Growing GAN和Face2Face)之间具有多大程度的泛化能力?
- RQ4在有限数据的基准数据集上,该方法是否显著优于现有最先进虚假检测模型?
- RQ5FDFtNet能否通过微调显著提升性能较差的基线模型(如SqueezeNet)的性能?
主要发现
- 在Deepfake数据集上,FDFtNet达到97.02%的最先进准确率,较之前最先进方法(Xception)高出0.6%。
- 在基于GAN的数据集(Progressive Growing GAN)上,FDFtNet达到90.29%的准确率,即使在低分辨率(64×64)下也表现出色。
- 消融实验表明,FTT模块贡献显著,将其加入Xception后准确率从94.56%提升至97.02%,提升2.46%。
- 所有基线模型,包括SqueezeNet(基线准确率50.00%),均得到显著改善,SqueezeNet经FDFtNet微调后准确率达到92.82%。
- 模型在不同数据集间泛化良好,将ResNetV2在Face2Face数据集上的准确率从58.83%提升至85.15%,并在该数据集上达到96.67%的准确率。
- FDFtNet在Deepfake数据集上达到99.37%的AUROC,表明其在区分真实与虚假图像方面具有高度可靠性。
更好的研究,从现在开始
从阅读论文到最终审阅,大幅缩短您的研究时间。
无需绑定信用卡
本解读由 AI 生成,并经人工编辑审核。