Automated Liar Recognition from Facial Expression Using Hybrid Feedforward Deep Neural Network And Convolutional Neural Network

The lie can be detected by the ability of people with no accurate constant face. Also, the enforcement of the personnel system to produce the psychological review based on the person who speaks lies is mentioned. The problem of managing the non-congruent gestures and the fidgeting for the excessive speech patterns and the information has been handled in the form of manufacturing is proposed. The portable network graphic images revealed the person who speaks and is telling a lie. This lie detection can be analyzed using PNG images. The process of analyzing the lie using the feed-forward deep neural network (ffDNN) and the image of GMM and the ResNet has been done. Using the ffDNN, the machine learning process based on precision-recall and the f-measures that can be calculated using the facial expression has been proposed. Also, the global optimum for identifying the feasible face who tells tie can be identified. Using ResNet, the classification of the images used as the backbone model and the field analyzed based on the AlexNet is proposed. This image classification results from analyzing the faces of all the people who are lier have been analyzed. The process of managing the faces of the database in the convolutional layer for the image classification and comparingng the 12 layers.The AlexNet is the layers, and the ResNet is the 10 layers. This layer manages, and the rest of the 112 layers have been proposed using the ImageNet. Using this calculation, the factor for applying the 3D geometric morphometric has been presented by machine learning has been submitted. This is how the liar can be detected.