arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2301.02145cs.CV
收藏

基于集成堆叠的域泛化方法用于人脸呈现攻击检测

Domain Generalization via Ensemble Stacking for Face Presentation Attack Detection

Aalto University(阿尔托大学);University of Oulu(奥卢大学)

Usman Muhammad, Jorma Laaksonen, Djamila Romaissa Beddiar, Mourad Oussalah

AI总结:

本文针对人脸呈现攻击检测的跨域泛化难题,提出结合alpha合成与视频蒸馏生成合成数据、并利用堆叠集成学习训练基模型与元模型的方法,在多个基准数据集上取得较低HTER。

摘要:

人脸呈现攻击检测(PAD)在保护人脸识别系统免受欺骗攻击方面发挥着关键作用。尽管在人脸PAD方法设计上已取得很大进展,但开发一个能够良好泛化到未见测试域的模型仍然是一个重大挑战。此外,由于欺骗攻击类型不同,创建具有足够数量样本的数据集以训练深度神经网络是一项费时费力的任务。本工作提出了一种综合解决方案,结合合成数据生成与深度集成学习来增强人脸PAD的泛化能力。具体而言,通过使用alpha合成和视频蒸馏将静态图像与时空编码图像混合来生成合成数据。通过这种方式,我们利用不同的alpha值模拟运动模糊,从而生成多样化的合成数据子集,有助于构建更丰富的训练集。此外,使用堆叠集成学习在每个合成数据子集上训练多个基模型。这使得模型能够从不同的合成子集中学习互补的特征和表示。由基模型生成的元特征被用作一个称为元模型的新模型的输入。元模型结合来自基模型的预测,利用它们的互补信息更好地处理未见目标域并提升整体性能。在四个数据集上的实验结果表明,在三个基准数据集上取得了较低的半总错误率(HTER):CASIA-MFSD(8.92%)、MSU-MFSD(4.81%)和OULU-NPU(6.70%)。该方法通过利用大规模合成数据和元模型,展示了推进呈现攻击检测的潜力。

英文摘要:

Face Presentation Attack Detection (PAD) plays a pivotal role in securing face recognition systems against spoofing attacks. Although great progress has been made in designing face PAD methods, developing a model that can generalize well to unseen test domains remains a significant challenge. Moreover, due to different types of spoofing attacks, creating a dataset with a sufficient number of samples for training deep neural networks is a laborious task. This work proposes a comprehensive solution that combines synthetic data generation and deep ensemble learning to enhance the generalization capabilities of face PAD. Specifically, synthetic data is generated by blending a static image with spatiotemporal encoded images using alpha composition and video distillation. This way, we simulate motion blur with varying alpha values, thereby generating diverse subsets of synthetic data that contribute to a more enriched training set. Furthermore, multiple base models are trained on each subset of synthetic data using stacked ensemble learning. This allows the models to learn complementary features and representations from different synthetic subsets. The meta-features generated by the base models are used as input to a new model called the meta-model. The latter combines the predictions from the base models, leveraging their complementary information to better handle unseen target domains and enhance the overall performance. Experimental results on four datasets demonstrate low half total error rates (HTERs) on three benchmark datasets: CASIA-MFSD (8.92%), MSU-MFSD (4.81%), and OULU-NPU (6.70%). The approach shows potential for advancing presentation attack detection by utilizing large-scale synthetic data and the meta-model.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑