Frontiers in Signal Processing

Download PDF (658.4 KB) PP. 81 - 87 Pub. Date: October 1, 2020

DOI: 10.22606/fsp.2020.44001

• Hao Gao
  College of Electrical & Information Engineering, Southwest Minzu University, Chengdu 610041, China
• Bo Ma^*
  College of Electrical & Information Engineering, Southwest Minzu University, Chengdu 610041, China

With the development of deep learning, even important progress has been made in the field of image classification and recognition. But facial expression recognition still faces many problems. This article is an experiment on the FER2013 dataset, the purpose is to get the facial expression attributes from the facial image. Because the pictures in this dataset have low resolution, and some pictures have no faces at all. This reduces the accuracy of facial expression recognition. In this paper, we propose a robust improved model. In this model, we introduce attention mechanism and separable convolution to improve the extraction of image features, and use data argumentation techniques to enhance the generalization ability of the model. The model obtained 65.2% test set accuracy on the FER2013 dataset.

attention mechanism, separable convolution, FER2013

[1] Zhiding Yu & Cha Zhang. (2015). Image based Static Facial Expression Recognition with Multiple Deep Network Learning. Proceedings of the 2015 ACM on International Conference on Multimodal Interaction. 435- 442.

[2] Abhinav Dhall, Roland Goecke, Simon Lucey, and Tom Gedeon. (2011).Static Facial Expressions in Tough Conditions: Data, Evaluation Protocol And Benchmark, First IEEE International Workshop on Benchmarking Facial Image Analysis Technologies BeFIT, IEEE International Confer-ence on Computer Vision ICCV2011, Barcelona, Spain, 6-13 November 2011.

[3] Ebrahimi Kahou, S., Michalski, V., Konda, K., Memisevic, R., and Pal,C. (2015). Recurrent neural networks for emotion recognition in video.Proceedings of the 2015 ACM on International Conference on Multimodal Interaction. 467-474. ACM.

[4] S. Zhou, Y. Liang, J. Wan, and S. Z. Li, “Facial expression recognition based on multi-scale cnns,” in Biometric Recognition, pp. 503–510,Springer, 2016.

[5] P. Carrier and A. Courville, “Challenges in representation learning:Facial expression recognition challenge.” https://goo.gl/kVzT48, 2013.

[6] Y. Tang, “Deep learning using linear support vector machines,” in Workshop on Challenges in Representation Learning, International Conference on Machine Learning (ICML), 2013.

[7] Yuan Tang. 2016. TF.Learn: TensorFlow’s High-level Module for Distributed Machine Learning. ArXiv preprint arXiv:1612.04251 (2016).

[8] F. Chollet. Xception: Deep learning with depthwise separable convolutions. arXiv preprint arXiv:1610.02357v2, 2016.

[9] Zhao M, Zhong S, Fu X, et al. Deep Residual Shrinkage Networks for Fault Diagnosis[J]. IEEE Transactions on Industrial Informatics, 2020, PP(99):1-1.

[10] Wan L, Zeiler M, Zhang S, et al.: “Regularization of neural networks using dropconnect,” Proceedings of the 30th International Conference on Machine Learning (ICML-13), 2013, pp.1058–1066.

[11] S. Ioffe and C. Szegedy. Batch normalization: Accelerating deep network raining by reducing internal covariate shift. In ICML, 2015.

[12] Hu J, Shen L, Albanie S, et al. Squeeze-and-Excitation Networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, PP(99).

[13] Zhai Y, Liu J, Zeng J, et al. Deep convolutional neural network for facial expression recognition[J]. 2017.

[14] Tumen V, Soylemez O F, Ergen B. Facial emotion recognition on a dataset using convolutional neural network[C]// 2017 International Artificial Intelligence and Data Processing Symposium (IDAP). IEEE, 2017.

[15] Liu K, Zhang M, Pan Z. Facial Expression Recognition with CNN Ensemble[C]// International Conference on Cyberworlds. IEEE Computer Society, 2016.

[16] Gan Y. Facial Expression Recognition Using Convolutional Neural Network[C]// the 2nd International Conference. 2018.