Face Gender And Age Recognition Based Convoiutioral Neural Network

In recent years,with the increasing use of neural networks in many fields,face image research has also received more and more attention.Recognition of age and gender based on face images has become one of the hot topics in current artificial intelligence research.Aiming at the problem that the single convolutional neural network does not recognize the face image's gender and age quickly and accurately,and is limited by the constraint image data set,two improved models based on convolutional neural network are proposed for the age and gender of the face image.The main contents of this paper are as follows:(1)A hybrid model of Convolutional Neural Network(CNN),Squeeze-Excitition Network(SENet),and Extreme Learning Machine(ELM)is proposed.The convolutional layer in the model is used to extract facial features from the face image.Furthermore,the SENet layer is used to optimize the features extracted by the convolutional layer.Finally,the Extreme Learning Machine(EM-ELM)is used as a classifier to carry out age and gender recognition from facial image.Compared with existing popular models,the proposed CNN-SE-ELM hybrid model uses the CNN + SENet architecture to enable it to extract more representative and optimal feature maps from facial images,while EM-ELM's extremely fast calculation makes the model faster and more efficient.(2)A hybrid model that combines the Oxford University Visual Geometry Group(VGG)and Squeeze-Excitition Network(SENet)is proposed.The VGGNet model is simplified to reduce the number of fully connected layers.At the same time,the SENet module is embedded.Finally,the classification loss function is improved to achieve age and gender classification of facial images.Compared with the existing popular models,the proposed VGG-SENet hybrid model uses deep convolution and SENet architecture to optimize the features extracted by the convolution layer and improve the classification loss function,so that the model classifies the gender and age of the face image faster and more accurately.Based on unrestricted IMDB-WIKI face data set and Adience face data set,the above two models are evaluated by using accuracy,precision,recall,and f1-score.The experimental results show that the accuracy of gender and age classification of CNN-SE-ELM model is96%,95% and 67%,65%,and the accuracy of sex and age classification of VGG-SENet model are 97%,96%,and 76%,74%.Compared with other related models,it is proved that the proposed CNN-SE-ELM and VGG-SENet models have higher recognition accuracy andspeed for gender and age classification of face images.