Since the distinction between two expressions is fairly vague, usually a subtle change in one part of the human face is enough to change a facial expression. Most of the existing facial expression recognition algorithms are not robust enough because they rely on general facial features or algorithms without considering differences between facial expression and facial identity. In this paper, we propose a person-independent recognition method based on Wasserstein generative adversarial networks for micro-facial expressions, where a facial expression recognition network and a facial identity recognition network are established to improve the accuracy and robustness of facial expression recognition via inhibition of intra-class variation. Extensive experimental results demonstrate that 90% average recognition accuracy of facial expression has been reached on a mixed dataset composed of CK+, Multi-PIE, and JAFFE. Moreover, our method achieves 96% accuracy of person-independent recognition on CK+. A 4.5% performance gain is achieved with the novel identity-inhibited expression feature. The proposed algorithm in this paper has been successfully applied to Haikang Visual Integrated Management Platform (iVMS-8700). At present, it runs well and can effectively recognize facial expressions.

中文翻译：

---

基于改进Wasserstein生成对抗网络结合身份感知的人脸表情识别方法

由于两种表情之间的区别相当模糊，通常人脸某一部分的细微变化就足以改变面部表情。现有的大多数面部表情识别算法都不够健壮，因为它们依赖于一般的面部特征或算法，而没有考虑面部表情和面部身份之间的差异。在本文中，我们提出了一种基于 Wasserstein 生成对抗网络的微面部表情识别方法，其中建立面部表情识别网络和面部身份识别网络，通过抑制来提高面部表情识别的准确性和鲁棒性。类内变异。大量实验结果表明，在由 CK+、Multi-PIE 和 JAFFE 组成的混合数据集上，面部表情的平均识别准确率已达到 90%。此外，我们的方法在 CK+ 上实现了 96% 的独立于人的识别准确率。新的身份抑制表达功能实现了 4.5% 的性能提升。本文提出的算法已成功应用于海康可视化综合管理平台（iVMS-8700）。目前，它运行良好，可以有效识别面部表情。本文提出的算法已成功应用于海康可视化综合管理平台（iVMS-8700）。目前，它运行良好，可以有效识别面部表情。本文提出的算法已成功应用于海康可视化综合管理平台（iVMS-8700）。目前，它运行良好，可以有效识别面部表情。