Article Details

Abstract

In order to improve the performance of facial expression recognition in unconstrained environments, a two-stage feature fusion expression recognition deep Convolutional Neural Network framework with embedded Attention Mechanism is studied and designed. This network framework designs and introduces multiple attention modules aimed at accurately extracting expression feature information of local image positions. Meanwhile, by constructing densely connected residual blocks, the quality of feature extraction is effectively improved and the stability of the network is enhanced. On this basis, the local features are fused with the global features extracted by the multi-scale module to obtain more discriminative expression features. The experimental results show that the proposed method exhibits good expression recognition performance on the RAF-DB dataset.

Keywords

expression recognition; Attention Mechanism; local feature; feature fusion

References

1. [1] Jiang Bin, Zhong Rui, Zhang Qiuwen, etc A review of non positive expression recognition using deep learning methods [J]. Computer Engineering and Applications, 2021, 57 (8): 48-61
2. [2] BARENTINE C，MCNAY A，PFAFFENBICHLER R，et al. A VR Teleoperation Suite with Manipulation Assist [C]//Companion of the 2021 ACM/IEEE International Conference on Human-robot Interaction.Boulder：ACM，2021：442-446.
3. [3] VERMA B，CHOUDHARY A. A Framework for Driver Emotion Recognition Using Deep Learning and Grassmann Manifolds [C]//2018 21st International Conference on Intelligent Transportation Systems（ITSC）.Maui：IEEE，2018：1421-1426.
4. [4] LI T H，DU C F，NAREN T Y，et al. Using Feature Points and Angles between them to Recognize Facial Expression by a Neural Network Approach [J].IET Image Processing，2018，12（11）：1951-1955.
5. [5] Jiang Bin, Cui Xiaomei, Jiang Hongbin, etc New progress of lightweight networks in facial expression recognition [J]. Computer Application Research, 2024, 41 (3): 663-670
6. [6] JIANG M，YIN S L. Facial Expression Recognition Based on Convolutional Block Attention Module and Multi-feature Fusion [J].International Journal of Computational Vision and Robotics，2023，13（1）：21-37.
7. [7] WANG K，PENG X J，YANG J F，et al. Region Attention Networks for Pose and Occlusion Robust Facial Expression Recognition [J].IEEE Transactions on Image Processing，2020，29：4057-4069.
8. [8] SZEGEDY C，LIU W，JIA Y Q，et al. Going Deeper with Convolutions [J/OL].arXiv:1409.4842 [cs.CV].（2014-09-17）.https://arxiv.org/abs/1409.4842.
9. [9] LI S，DENG W H，DU J P. Reliable Crowdsourcing and Deep Locality-preserving Learning for Expression Recognition in the Wild [C]//2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Honolulu：IEEE，2017：2584-2593.
10. [10] TENG J N，ZHANG D，ZOU W，et al. Typical Facial Expression Network Using a Facial Feature Decoupler and Spatial-temporal Learning [J].IEEE Transactions on Affective Computing，2023，14（2）：1125-1137 ：1125-1137.

Funding

Innovation Strong School Engineering Research Project of Guangdong University of Technology (2022GKJZK004); Key Discipline Project of Artificial Intelligence at Guangdong University of Technology (2024KDZK001); Guangdong University of Technology Experimental Teaching Demonstration Center Project (SFZX202402)

About the authors

Jiang Tao (October 1983-), male, Han ethnicity, from Anfu, Jiangxi, associate professor, master's degree, research direction: deep learning, computer vision; Li Chuzhen (1989.10-), female, Han ethnicity, from Chaozhou, Guangdong, associate professor, master's degree, research direction: deep learning.