Method for detecting face symmetry and anomaly based on real-time face detection

Abstract

The invention relates to a method for detecting face symmetry and anomaly based on real-time face detection. The method includes a training stage and a testing stage. The training stage includes the following steps that: a deep convolutional neural network model is established; the image data of the main parts of the face of a user are acquired; and an optimal classification strategy is provided according to the image data given by the user. The testing stage includes the following steps that: the facial movement of the user is monitored in real time; the image data of the main parts of the face of the user are acquired; the images of the main parts of the face of the user are processed through the deep convolutional neural network model, and deep convolutional features are extracted; a deep convolutional feature-based dichotomy problem is solved, state identification is carried out on the main parts of the face of the user, and then real-time symmetry detection and anomaly detection are carried out on the main parts; and the eye state time sequence of the user is recorded, and the sampling frequency of a monitoring camera is adjusted in real time. With the method of the invention adopted, real-time face detection on face symmetry and anomaly can be carried out accurately under conditions of different lighting conditions and different users. The method has high stability and high universality.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a method for detecting the symmetry and abnormality of a human face based on real-time face monitoring. Background technique [0002] In recent years, face and facial expression recognition has attracted more and more researchers' attention. Face detection refers to the process of finding all face information in video sequences or images, and determining the size, position, trajectory, and posture of the face, and further extracting features such as eyes and lips on the face. Face detection and recognition are classic problems in the field of pattern recognition. It designs knowledge and technologies in the fields of image processing, pattern recognition, and physiology. The research originated from the work of French scientist Alton in the 19th century. With the development of computer science, face detection gradually became the focus of research after the 1990s. In...

AI Technical Summary

Problems solved by technology

[0016] Steps in the test phase: monitor the user's facial activities in real time, collect the image data of the main parts of the user's face, process the images of the main parts of the user's face through the deep convolutional neural network model, and extract the deep convolution features; solve the binary classification based on the deep convolution features The problem is to identify the main parts of the user's face and then perform real-time symmetry detection and anomaly detection on the main parts; record the user's eye state time series, and adjust the sampling frequency of the surveillance camera in real time

Images