Abstract: Face recognition technology is an important way of identity authentication. The purpose of this paper was to design an algorithm to identify whether the identity witness and the person to be inspected were the same passengers. Face recognition algorithm based on convolutional neural network was studied. The algorithm was designed by calculating the Euclidean distance between face features after face detection, and the final accuracy was 95%. The results show that Faster R-CNN algorithm can detect face accurately, VGG-Net can extract face feature values well.