Human face key point detection method and system based on dynamic cascade regression

Abstract

The invention discloses a human face key point detection method and system based on dynamic cascade regression, and the method comprises the steps: obtaining a to-be-detected human face image, inputting the to-be-detected human face image into a global estimation network, and outputting a rough initial value of a human face key point position; and inputting the rough initial value of the face keypoint position into a local regression network with a termination criterion to carry out detailed iterative regression, and outputting a face key point detection result of the to-be-detected face picture. According to the invention, rough face key point position estimation can be provided by using a method based on a direct shape regression model, and then a cascade regression model with a termination criterion is used. Starting from rough face key point position estimation, detailed iterative regression is carried out, and the iteration step number is controlled by using the termination criterion, so that the detection precision of the face key points is effectively improved.

Description

technical field [0001] The invention relates to the technical field of human face key point processing, in particular to a method and system for detecting human face key points based on dynamic cascade regression. Background technique [0002] As a basic computer vision processing task, face key point detection not only helps face recognition tasks, but also lays the foundation for expression recognition tasks. [0003] At present, the more traditional face key point detection methods based on cascaded regression mainly use artificial features to drive the regression process, which is easy to fall into local optimum. In order to overcome the shortcomings of artificial features, most of the work based on cascade regression in recent years uses neural networks to complete the process of feature extraction and face key point regression. Simply, this type of method can be divided into two categories, one is to use the image blocks extracted around the key points as the input of...

AI Technical Summary

Problems solved by technology

[0004] It can be seen from this that the current face key point detection method based on the cascade regression model is very sensitive to the provided initial face key point position, especially for the method based on the local regression model, when the provided initial face key point When the position is far from the real position, these methods are particularly prone to fall into local optimum, making the key point detection accuracy relatively poor

In addition, the current face key point detection method based on the cascaded regression model basically uses a fixed number of regression iteration steps, and cannot increase or decrease the number of iteration steps according to the actual situation.

Images