基于正反残差块的人脸表情识别算法  

Facial Expression Recognition Algorithm Based on Residual and Inverted Residual Blocks

作　　者：俞成 姚瑶 张青山 宋晨阳 姜代红[1] YU Cheng;YAO Yao;ZHANG Qing-shan;SONG Chen-yang;JIANG Dai-hong(Department of Information and Electrical Engineering,Xuzhou Institute of Technology,Xuzhou Jiangsu 221008,China)

机构地区：[1]徐州工程学院信电工程学院,江苏徐州221018

出　　处：《软件》2020年第5期156-159,共4页Software

摘　　要：针对传统人脸表情识别算法鲁棒性低,传统表情识别特别依赖人工设计的特征,提出一种改进的人脸表情识别算法。该算法在深度学习中的残差块与反残差块算法基础上,将两种算法融合,利用两种神经块结合起来的深度学习模型,同时配合数据增强、Dropout等技术。实验结果表明,经过优化的算法,在正确率上比原先有一定的提升,在训练稳定性上也有较好的表现,同时具有了更好的鲁棒性。In order to solve the following problems,including poor robustness and requirement of lots of manpower,an improved model based on residual block and inverted residual block was proposed.By combining the two technologies along with data augmentation and Dropout,this model is built.The experimental results show that the improved algorithm works well compared to other models.

关 键 词：残差块 反残差块 人脸表情识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]