Facial feature point positioning method, device, apparatus and medium based on cascade regression

Abstract

As that invention is applicable to the field of computer technology, A facial feature point localization method based on cascade regression is provided, a device, an apparatus and a medium are provided. The method comprises: Through the multi-scale convolution neural network in the trained cascade regression model, predicting a feature point position set of a face image to be located and extracting a global feature of the image, according to the global feature, modifying the feature point position set of the face image to be located by each refinement layer in the cascade regression model, according to the corrected feature point position set, the final position of all feature points on the face image to be located is determined, Among them, the cascade regression model is based on the Jacobian matrix in-depth optimization strategy training, Each refinement layer includes a local feature extractor, a feature fusion device and a regressor, which improves the effect of image texture information utilization and model training in the process of facial feature point positioning, and further improves the accuracy and accuracy of facial feature point positioning.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a method, device, equipment and medium for locating facial feature points based on cascade regression. Background technique [0002] Facial feature point location is an important part of face animation generation based on face images. Usually, a reliable and fast face detection algorithm first gives the face area as input, and then a set of presets in the face area Locate the defined facial feature points, such as eyebrows, eye pupils, nose tip, mouth corners, etc. At present, the facial feature point localization methods are mainly divided into two types of models: the generative model represented by the classic algorithm active appearance model, and the discriminative model represented by the cascaded regression method (such as the supervised descent method) that has emerged in recent years. The generative model often fails when locating the feature points...

AI Technical Summary

Problems solved by technology

[0004] The object of the present invention is to provide a facial feature point location method based on cascade regression, device, equipment and storage medium, aiming to solve the problem that the facial feature point location method based on cascade regression in the prior art cannot make good use of human The texture information of the face image and the model training effect are not good, resulting in the problem of low accuracy and poor effect of facial feature point positioning

Images