Sparse deep feature learning for facial expression recognition

Weicheng Xie Xi Jia Linlin Shen Meng Yang

Computer Vision Institute, School of Computer Science & Software Engineering, Shenzhen University, China
Guangdong Key Laboratory of Intelligent Information Processing, China
Shenzhen Institute of Artificial Intelligence and Robotics for Society, Sun Yat-sen University, China
School of Data and Computer Science, Sun Yat-sen University, China

Abstract

While weight sparseness-based regularization has been used to learn better deep features for image recognition problems, it introduced a large number of variables for optimization and can easily con- verge to a local optimum. The L2-norm regularization proposed for face recognition reduces the impact of the noisy information, while expression information is also suppressed during the regularization. A feature sparseness-based regularization that learns deep features with better generalization capability is proposed in this paper. The regularization is integrated into the loss function and optimized with a deep metric learning framework. Through a toy example, it is showed that a simple network with the proposed sparseness outperforms the one with the L2-norm regularization. Furthermore, the proposed approach achieved competitive performances on four publicly available datasets, i.e., FER2013, CK + , Oulu-CASIA and MMI. The state-of-the-art cross-database performances also justify the generalization capability of the proposed approach

Visualization

The ‘disgust’ expression of different persons in different databases

The data preprocessing

The hidden unit, weight and the proposed feature sparseness

The structures of two networks

The fine tuning of models for different databases

Example images of FER2013, CK + , Oulu-CASIA and MMI

The comparison of the L2-norm regularization and the L2 sparseness

The average ROC curves and the corresponding values of AUC

The validation and testing accuracies


@article{xie2019sparse,
    title={Sparse deep feature learning for facial expression recognition},
    author={Xie, Weicheng and Jia, Xi and Shen, Linlin and Yang, Meng},
    journal={Pattern Recognition},
    volume={96},
    pages={106966},
    year={2019},
    publisher={Elsevier}
    }