Feature super-resolution based Facial Expression Recognition for multi-scale low-resolution images
Feature super-resolution based Facial Expression Recognition for multi-scale low-resolution images
复制标题
基于特征超分辨率的多尺度低分辨率图像面部表情识别
DOI：
10.1016/j.knosys.2021.107678
复制
发表时间：
2021-11
影响因子：
8.8
通讯作者：
Qinghua Zheng
中科院分区：
文献类型：
--
作者：
Fang Nan;Wei Jing;Feng Tian;Jizhong Zhang;Kuo-Ming Chao;Zhenxin Hong;Qinghua Zheng
各种低分辨率图像的面部表情识别是人群场景（车站、教室等）分析应用中的一项重要任务和需求。由于分辨率降低导致的判别特征损失，将各种低分辨率人脸图像分类到正确的类别中仍然是一项具有挑战性的任务。在这项工作中，我们提出了一种新的基于生成对抗网络的特征级超分辨率鲁棒面部表情识别（FSR-FER）方法，该方法可以在不恢复高分辨率面部图像的情况下减少隐私泄露的机会。特别地，使用预训练的FER模型作为特征提取器，并使用从低分辨率图像和相应的高分辨率图像中提取的特征训练生成器网络G和判别器网络D。生成器网络G试图通过使低分辨率图像的特征更接近相应的高分辨率图像的特征，将低分辨率图像的特征转化为更具判别性的特征。为了获得更好的分类性能，我们还提出了一种有效的基于固定FER模型计算的分类概率的分类感知损失重加权策略，使我们的模型更加关注容易出现误分类的样本。在真实世界情感面孔（RAF）数据库和野外静态面部表情（SFEW） 2.0数据集上的实验结果表明，我们的方法在单一模型下对各种下采样因素都取得了令人满意的结果，并且与分别使用图像超分辨率和表情识别的方法相比，在低分辨率图像上具有更好的性能。
Facial Expression Recognition (FER) for various low-resolution images is an important task and need in applications of analyzing crowd scenes (station, classroom, etc.). Due to the discriminative feature loss caused by reduced resolution, classifying various low-resolution facial images into the right category is still a challenging task. In this work, we proposed a novel generative adversarial network-based feature level super-resolution method for robust facial expression recognition (FSR-FER), which can reduce the chance of privacy leaking without restoring high-resolution facial images. In particular, a pre-trained FER model was employed as a feature extractor, and a generator network G and a discriminator network D are trained with features extracted from low-resolution and corresponding high-resolution images. Generator network G tries to transform features of low-resolution images to more discriminative ones by making them closer to the ones of corresponding high-resolution images. For better classification performance, we also proposed an effective classification-aware loss reweighting strategy based on the classification probability calculated by a fixed FER model to make our model focus more on samples that are prone to misclassification. Experimental results on the Real-World Affective Faces (RAF) Database and Static Facial Expressions in the Wild (SFEW) 2.0 dataset demonstrate that our method achieves satisfying results on various down-sample factors with a single model and has better performance on low-resolution images compared with methods using image super-resolution and expression recognition separately.
登录
查看更多内容
DOI：
10.1088/1742-6596/1168/2/022043
发表时间：
2019-02
期刊：
Journal of Physics: Conference Series
影响因子：
--
作者：
Jielong Tang;Xiaotian Zhou;Jiawei Zheng
通讯作者：
Jielong Tang;Xiaotian Zhou;Jiawei Zheng
DOI：
--
发表时间：
2012-11
期刊：
Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012)
影响因子：
--
作者：
Nikolas Hesse;Tobias Gehrig;Hua Gao;H. K. Ekenel
通讯作者：
Nikolas Hesse;Tobias Gehrig;Hua Gao;H. K. Ekenel
影响因子：
16
作者：
Xiangyu Zhao;Xiaodan Liang;Luoqi Liu;Teng Li;Yugang Han;N. Vasconcelos;Shuicheng Yan
通讯作者：
Xiangyu Zhao;Xiaodan Liang;Luoqi Liu;Teng Li;Yugang Han;N. Vasconcelos;Shuicheng Yan
DOI：
10.1007/11744047_45
发表时间：
2006-05
期刊：
--
影响因子：
--
作者：
Oncel Tuzel;F. Porikli;P. Meer
通讯作者：
Oncel Tuzel;F. Porikli;P. Meer
影响因子：
3.2
作者：
Dhall, Abhinav;Goecke, Roland;Gedeon, Tom
通讯作者：
Gedeon, Tom