Research On Face Detection Based On Two-layer Cascaded Convolutional Neural Network

The traditional methods of face detection is limited by the face multi-pose changes and the incomplete facial features,which leads to the poor detection effect.To solve the problem,this paper proposed a face detection method of two-layer cascaded convolutional neural network.Firstly,a two-layer convolutional neural network model is constructed.The first convolutional neural network model is used to extract the features of the face image roughly,and then the max pooling method is adopted to reduce the dimension of those features,so as to output multiple suspected face windows.Secondly,the face windows are used as the inputs of the second convolutional neural network model for fine feature extraction,and a new feature map is obtained by pool operation.Finally,classification and discrimination through full connection layer and output layer of the second convolutional neural network,the best detecion window is output by non-maxinum suppression method,and the whole process of face detection is completed.The experiment selected those images with face multi-pose changes and with the incomplete facial feature information in the FDDB face detection dataset to carry on the test,the two-layer convolutional neural network algorithm is evaluated in terms of true detection rate,false detection rate,miss detection rate and detection time.The experimental results show that two-layer convolutional neural network algorithm can improve the true detection rate while guaranteeing the detection speed of the algorithm compared with the current popular methods.This two-layer cascaded convolutional neural network for face detection can ensure a high detection rate under the complex condition of face multi-pose change or when the facial feature information is incomplete,the method has good robustness and generalization ability.There are 27 figures,3 tables and 58 references in this paper.