Transformer And Node-Compressed Dnn Based Dual-Path System For Manipulated Face Detection
Transformer And Node-Compressed Dnn Based Dual-Path System For Manipulated Face Detection
复制标题
DOI：
10.1109/icip42928.2021.9506222
复制
发表时间：
2021-09
期刊：
影响因子：
--
通讯作者：
Zhengbo Luo;S. Kamata;Zitang Sun
中科院分区：
文献类型：
--
作者：
Zhengbo Luo;S. Kamata;Zitang Sun
深度神经网络（dnn）广泛地促进了数据生成的发展；这些生成内容的质量达到了令人印象深刻的新水平。因此，被操纵的内容，尤其是面部操纵，是网络信息合法性日益受到关注的问题。目前大多数基于深度学习的方法依赖于卷积核采样的局部特征，缺乏全局知识。为了解决这个问题，我们提出了一种基于神经常微分方程（NODE）的神经网络和面部特征偏差变压器的双路径管道，从不同的角度处理视觉内容。此外，我们还采用了一种基于注意力引导增强的自集成，以提高系统的鲁棒性。大量的实验表明，我们的系统在视频级精度和AUC方面可以超越几种常用的方法，并且具有更好的可解释性。
Deep neural networks (DNNs) have extensively promoted data generation development; the quality of these generated content has achieved an impressive new level. Therefore, manipulated content, especially facial manipulation, is a growing concern for online information legitimacy. Most current deep learning-based methods depend on local features sampled by convolutional kernels and lack knowledge globally. To address the problem, we propose a dual-path pipeline using Neural Ordinary Differential Equations (NODE) based neural network and facial-feature biased transformer to deal with the visual content from a different view. The transformer path could link these landmarks in a long-range, moreover, we adopt an attention guided augmentation based self-ensemble for more robust performance. Extensive experiments show that our system could surpass several commonly used approaches in terms of video-level accuracy and AUC with better interpretability.