This article proposes a vision Transformer facial expression recognition algorithm based on residual attention network, aiming to improve the accuracy and efficiency of facial expression recognition. By combining the residual network and the attention mechanism, the feature extraction ability is enhanced, enabling the model to capture the subtle features of facial expressions more effectively. The improved Transformer performs extremely well in dealing with complex backgrounds and multi-scale targets, significantly improving the recognition performance. The experimental results show that this method has achieved excellent detection effects on multiple public datasets.

Yuexin Lin. Research on Vision Transformer Facial Expression Recognition Algorithm Based on Residual Attention Network. Academic Journal of Computing & Information Science (2025), Vol. 8, Issue 6: 13-18. https://doi.org/10.25236/AJCIS.2025.080602.

[1] Mehrabian A. Communication without words [J]. Communication Theory, 2008, 193-200.

[2] Chen Gang, Zhang Shiqing, Zhao Xiaoming. Expression Recognition of Video sequences using Transformer Network [J]. Journal of Image and Graphics, 2022, 27(10):3022-3030.

[3] Li N ,Huang Y ,Wang Z , et al.Enhanced Hybrid Vision Transformer with Multi-Scale Feature Integration and Patch Dropping for Facial Expression Recognition[J].Sensors,2024,24(13):4153.

[4] Wang Bin, LIU Tianpei, HUANG Mingliang. Multi-scale high-order reduced bilinear pooling for facial expression recognition[J]. Jiangsu Communications, 2024, 40(05): 98-103.

[5] Wang K, Peng X, Yang J, et al. Region attention networks for pose and occlusion robust facial expression recognition[J]. IEEE Transactions on Image Processing, 2020, 29: 4057-4069.

[6] Xue Zhichao, Yilihamu Yarmamat, Yan Tianxing. Facial Expression Recognition Based on Multi-scale Feature Fusion of MobileNetV3[J]. Electronic Measurement Technology, 2023, 46(08): 38-44. DOI: 10.19651/j.cnki.emt.2211334.