摘要
提出一种基于Adaboost算法对人脸图像进行情绪识别的方法。先将视频进行图像数据采集,再通过基于Haar特征值的自适应增强计算,即Adaboost计算检测人脸特征,将迁移机器学习技术运用到多任务的卷积式神经网络,然后利用卷积神经网络的情绪回归计算人脸表情的效价和唤醒度得分。不但可以解决对复杂背景图像的高误检率问题,而且还可以解决对多姿态单人脸图像的低检率问题。经仿真试验证明,该方法对单人脸、多人脸和复杂背景多人图像都有较好的测量效果,实用性较强。
This paper proposes a method for emotion recognition of face images based on the AdaBoost algorithm.The method uses the system to call the camera to sample the video image data,and uses the adaptive enhancement algorithm based on Haar feature values,namely,the AdaBoost algorithm to detect face features,uses transfer learning technology to train a multi-task convolutional neural network,and then uses the emotion regression of the convolutional neural network to calculate the valence and arousal scores of facial expressions.This method can simultaneously solve the problems of high false detection rate for complex background images and low detection rate for multi-pose face images.Simulation experiments show that this method has better detection results for single-face,multiple-face and multiple-person images with complex backgrounds,and has strong practicability.
作者
王燕
WANG Yan(Zhejiang Police Vocational Academy,Hangzhou,Zhejiang 310004,China)
出处
《杨凌职业技术学院学报》
2023年第1期10-13,共4页
Journal of Yangling Vocational & Technical College
作者简介
王燕(1976-),女,江西南昌人,副教授,硕士研究生,研究方向为人工智能技术,新一代互联网技术。
