Deep Convolutional Neural Networks

在过去的几年中，深度卷积神经网络（CNN）一直在推动人脸识别的前沿。然而，现有的通用CNN面部模型对于可变面部区域上的遮挡效果的概括性较差。受人类视觉系统显式忽略遮挡而只关注未遮挡的面部区域这一事实的启发，我们提出了一种遮罩学习策略来发现和丢弃损坏的特征元素以使其无法识别。首先使用创新设计的成对差分暹罗网络（PDSN），利用遮挡和无遮挡脸部对的顶级平移特征之间的差异来建立遮罩字典。该词典的每一项都记录了被遮挡的面部区域与损坏的要素元素之间的对应关系，称为“要素丢弃遮罩（FDM）”。当处理具有随机局部遮挡的面部图像时，我们通过组合相关的字典项来生成其FDM，然后将其与原始特征相乘以从识别中消除那些损坏的特征元素。综合的和现实的遮挡人脸数据集的综合实验表明，所提出的算法明显优于最新技术的系统。

在过去的几年里，深度卷积神经网络（CNN）一直在推动人脸识别的前沿。然而，现有的一般CNN面部模型对可变面部区域的遮挡效果不佳。受到人类视觉系统明显忽视遮挡的启发，只关注非封闭的面部 ar-eas，我们提出了一个面具学习策略，以从识别中查找和消除卡损坏的特征元素。口罩词典首先通过利用屏蔽和遮挡无遮挡面对的顶部conv特征之间的差异，使用创新设计的双向分体暹罗网络（PDSN）来建立的。此 dic-tionary 的每个项目都捕获了被屏蔽的外层区域和损坏的功能元素之间的通信，后者被命名为"功能丢弃掩码 "（FDM）。在处理带有随机部分遮挡的面部图像时，我们通过结合相关的字典项生成其 FDM，然后将其与原始功能乘以以消除识别中那些被腐蚀的特征元素。综合和逼真的遮挡面部数据集的综合实验表明，所提议的算法明显优于最先进的系统。

Deep Convolutional Neural Networks (CNNs) have been pushing the frontier of face recognition over past years. However, existing general CNN face models generalize poorly for occlusions on variable facial areas. Inspired by the fact that the human visual system explicitly ignores the occlusion and only focuses on the non-occluded facial ar- eas, we propose a mask learning strategy to find and dis- card corrupted feature elements from recognition. A mask dictionary is firstly established by exploiting the differences between the top conv features of occluded and occlusion- free face pairs using innovatively designed pairwise dif- ferential siamese network (PDSN). Each item of this dic- tionary captures the correspondence between occluded fa- cial areas and corrupted feature elements, which is named Feature Discarding Mask (FDM). When dealing with a face image with random partial occlusions, we generate its FDM by combining relevant dictionary items and then multiply it with the original features to eliminate those cor- rupted feature elements from recognition. Comprehensive experiments on both synthesized and realistic occluded face datasets show that the proposed algorithm significantly out- performs the state-of-the-art systems.<br>