Face image detection method and system, storage medium and equipment

Abstract

The invention provides a face image detection method and system, a storage medium and equipment. The method comprises the steps of extracting a face image in a to-be-processed image; obtaining the confidence coefficient of the thermodynamic diagram corresponding to each key point in the face image; and calculating a face key point score of each face image according to each confidence coefficient so as to judge the shielded degree of the face in the face image. According to the human face image detection method and system, the storage medium and the equipment, key point positioning and detection can be performed on the shielded human face data, and the target human face can be quickly and effectively recognized and found under the conditions of different light rays, different angles, different postures and the like.

Description

technical field [0001] The present invention relates to the technical field of face detection, in particular to a face image detection method, system, storage medium and equipment. Background technique [0002] Face key point detection is to predict the coordinates of a series of predefined key points such as facial features and facial contours based on a given face, so as to carry out subsequent applications such as face alignment, recognition, and expression analysis. [0003] Traditional facial key point detection methods mainly include classic algorithms such as active appearance model, active shape model and cascaded regression model. With the introduction and development of deep learning, some methods based on deep convolutional neural network have emerged in recent years for face key point detection. At present, the most mainstream method is the method of using coordinate regression and heat map regression. Among them, the coordinate regression model The main idea is...

AI Technical Summary

Problems solved by technology

Its advantage is that it does not require too much data post-processing, and the network can directly return the coordinates of the output key points to achieve end-to-end training and forward reasoning, but its accuracy is inferior to that of the heat map regression method.

Moreover, since the method of coordinate regression is limited by the geometric positional relationship of key points of the face, the prediction result depends on the face detector, and there are two problems as follows: one is when the face detector detects a face, if the person The face is a side face, and the key points obtained by the coordinate regression method will predict the key points of the occluded part according to the position of the visible key points, combined with the geometric contour constraints of the face, so that the forced output of a specified number of key points is not conducive to subsequent Face alignment and recognition; the second is for images that do have faces, limited by the accuracy of the face detector algorithm, if no visible face is detected in the picture, then the key point of the corresponding coordinate regression is Unable to obtain, thus losing visible and useful face information

Images