面向智慧养老的改进YOLOv7人脸识别算法

Improved YOLOv7 Face Recognition Algorithm for Smart Pension

针对目前社会老龄化趋势,为适应养老领域发展需求,提出面向智慧养老的改进YOLOv7人脸识别算法。首先,提出多尺度信息输入模块,提取图像的全局信息,提高信息利用率;其次,总结归纳老年人脸特征,提出全局自适应特征提取模块,结合注意力机制改进主干网络和检测头;最后,通过迁移学习方法训练网络,并加入多项式损失策略分配特征权重,同时不断调试参数来提高网络识别能力。实验结果表明,所提网络在老年人脸数据上的精度和召回率分别为95.26%和91.57%,并且相比于原YOLOv7的网络参数量下降了5.4%。

To meet the current aging trend of society and meet the development needs of the field of old-age care,an improved YOLOv7 face recognition algorithm for smart pension is proposed.Firstly,a multi-scale information input module is proposed to extract the global information of the image and improve the information utilization rate.Secondly,the face features of the elderly are summarized,and a global adaptive feature extraction module is proposed,which combines attention mechanism to improve the backbone network and detection head.Finally,the network is trained by transfer learning method,the feature weights are assigned by polynomial loss strategy,and the parameters are continuously debugged to improve the network identification ability.The experimental results show that the accuracy and recall of the proposed network on the old face data set reach 95.26%and 91.57%respectively,and the network parameters are reduced by 5.4%compared with the original YOLOv7.