重庆理工大学学报(自然科学) ›› 2023, Vol. 37 ›› Issue (1): 177-185.

• 信息·计算机 • 上一篇    下一篇

深度宽残差网络注意力机制的人脸表情识别

倪锦园,张建勋,张馨月   

  1. .重庆理工大学 计算机科学与工程学院,重庆 400054; 2.东北大学悉尼智能科技学院,河北 秦皇岛 06600
  • 出版日期:2023-02-16 发布日期:2023-02-16
  • 作者简介:倪锦园,男,硕士研究生,主要从事数字图像处理与分析、人脸识别研究,Email:1833474277@qq.com;通讯作者 张建勋,男,博士,教授,主要从事数字图像处理与分析、实时计算机图形学研究,Email:466908695@qq.com

Facial expression recognition based on the attention mechanism of deep and wide residual networks

  • Online:2023-02-16 Published:2023-02-16

摘要: 针对自然状态下的人脸表情识别精度不高,易受噪声等因素的影响,提出了改进的 深度宽残差网络并融合注意力机制的人脸表情识别方法。通过拓宽残差单元的通道数形成一 种宽残差模块结构,有效减缓了网络层数过多造成梯度消失的问题;为提高网络对面部特征的 表示能力,引入了压缩和自适应校正网络模块;为减缓模型过拟合的现象,改进了残差单元的顺 序;通过改进的随机擦除方法对原始图像进行处理,进一步加强了模型的泛化能力。实验结果 表明:模型在 fer2013、ck+数据集和 JAFFE数据集上的准确率分别为 72.49%、99.29%和 9487%,与其他方法相比,所提模型在识别准确性上有较大提升,同时具有较好的鲁棒性。

关键词: 表情识别, 宽残差网络, 过拟合, 随机擦除

Abstract: Aiming at a low accuracy of facial expression recognition and the susceptibility to noise and other factors under natural conditions, this paper proposes a facial expression recognition method that incorporates an improved deep and wide residual network with attention mechanism. A wide residual module structure is formed by increasing the channel number ofresidual units, which effectively alleviates the problem of gradient disappearance caused by excessive network layers.In order to understand facial expressionfornetworks, a compressed and adaptive correction network module is introduced.In order to alleviateover-fitting of the model, the order of the residual units is improved.The original imagesare processed through the improved random erasure method to further strengthen the generalization ability of the model. Experimental results show that the accuracy of the model on fer2013, ck+ data set and JAFFE data set are 72.49%, 99.29% and 94.87%respectively.Compared with other methods, the model proposed in this article has a much higher recognition accuracy and, at the same time,a better robustness.

中图分类号: 

  • TP311.11