Advances in International Computer Science
Advances in International Computer Science. 2022; 2: (2) ; 10.12208/j. aics.20220017 .
总浏览量: 1138
浙江树人大学信息科技学院 浙江杭州
*通讯作者： 马铭骏,单位：浙江树人大学信息科技学院 浙江杭州；
真实的人机交互场景下，人脸面部图像会受到光照等因素影响，从而降低面部表情识别准确率。针对该问题，提出了一种基于光照归一化和ResNet18的优化模型。采用直方图均衡化和线性变换加权求和的光照归一化方法对原图进行亮度平衡，并利用预训练的ResNet18网络提取面部特征；使用Softmax函数对面部表情结果进行预测。实验结果表明，该网络模型在RAF-DB和FERPlus上分别取得87.03%和87.46%识别准确率。
In the real human-computer interaction scene, facial images of human faces are affected by factors such as lighting, which reduces the accuracy of facial expression recognition. In view of the problem, a optimization model based on improved illumination normalization and ResNet18 was proposed. The illumination normalization method of histogram equalization and linear transformation weighted summation is used to balance the brightness of the original image, and extract facial features using the pre-trained ResNet18 network; Softmax function was used to predict the facial expression results. The experimental results show that the network model achieves 87.03% and 87.46% recognition accuracy on RAF-DB and FERPlus datasets.
[1] Charles D. The Expression of the Emotions in Man and Animals [M]. New York City: Oxford University Press, 2002.
[2] Shan Li, Weihong Deng. Deep facial expression recognition: A survey[J/OL]. IEEE Trans on Affective Computing, 2020, 2020(2): Article ID 2981446. (2020-03-17)[2022-01-18]. http://doi.org/10.1109/taffc.2020.2981446
[3] 国家统计局, 国务院第七次全国人口普查领导小组办公室. 第七次全国人口普查公报~([1])(第五号) [N]. 中国信息报, 2021-05-12(002).
[4] 李勇, 林小竹, 蒋梦莹. 基于跨连接LeNet-5网络的面部表情识别 [J]. 自动化学报, 2018, 44(01): 176-182. (Li Yong, Lin Xiaozu, Jiang Mengying. Facial Expression Recognition with Cross-connect LeNet-5 Network [J]. Acta Automatica Sinica, 2018, 44(01): 176-182)
[5] SHAO, Jie; QIAN, Yongsheng. Three convolutional neural network models for facial expression recognition in the wild. Neurocomputing, 2019, 355: 82-92.
[6] Yong Li, Jiabei Zeng, Shiguang Shan, et al. Occlusion Aware Facial Expression Recognition Using CNN With Attention Mechanism [J]. IEEE Transactions on Image Processing, 2018, 28(5): 2439-2450.
[7] Kai Wang, Xiaojiang Peng, Jianfei Yang, et al. Region Attention Networks for Pose and Occlusion Robust Facial Expression Recognition [J]. IEEE Transactions on Image Processing, 2020(29): 4057-4069.
[8] Ioannis A K, George T, Georgios E, et al. 3D-2D Face Recognition with Pose and Illumination Normalization [J]. Computer Vision and Image Understanding, 2017(154): 137-151.
[9] Wuming Zhang, Xi Zhao, Jean-Marie M, et al. Improving Shadow Suppression for Illumination Robust Face Recognition [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 41(3): 611-624.
[10] Chieh-Ming K, Shang-Hong L, Michel S. A Compact Deep Learning Model for Robust Facial Expression Recognition [C]// Proc of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops. 2018: 2121-2129.
[11] Haiping Wu, Bin Xiao, Noel C, et al. CvT: Introducing Convolutions to Vision Transformers [C]// Proc of the IEEE/CVF International Conference on Computer Vision (ICCV). 2021: 22-31.
[12] Shan Li, Weihong Deng, JunPing Du. Reliable Crowdsourcing and Deep Locality-Preserving Learning for Expression Recognition in the Wild [C]// Proc of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017: 2852-2861.
[13] Emad B, Cha Zhang, Cristian C F, et al. Training deep networks for facial expression recognition with crowd-sourced label distribution [C]// Proc of the 18th ACM International Conference on Multimodal Interaction. 2016:279-283.
[14] Kaipeng Zhang, Zhanpeng Zhang, Zhifeng Li, et al. Joint Face Detection and Alignment Using Multitask Cascaded Convolutional Networks [J]. IEEE signal processing letters, 2016, 23(10): 1499-1503.