Method and device for obtaining feature points of human face

Abstract

The invention discloses a method and device for obtaining the feature points of a human face. The method comprises the steps: obtaining a human face after normalization processing; inputting the humanface into a convolution neural network for processing, wherein the convolution neural network comprises at least one convolution layer, at least one pooling layer, at least one local response normalization layer and at least one total connection layer, and the convolution layer is used for the convolution processing according a convolution core, the pooling layer is used for simplifying the dataof the convolution layer, and the configuration of the convolution neural network is correlated with the expected number of feature points of the human face, and is obtained through training via a predetermined training set; obtaining a plurality of feature points, processed by the convolution neural network, of the human face in the human face image. The method solves technical problems that a method for obtaining the feature points of the human face based on the convolution neural network in the prior art is large in calculation burden, and is longer in training time.

Description

technical field [0001] The present invention relates to the field of images, in particular to a method and device for acquiring facial feature points. Background technique [0002] Facial feature point acquisition is to automatically locate the key feature points of the face based on the input face image. According to different application requirements, the number of key feature points is also different: the minimum is 5 feature points, including eyes, nose tip and mouth corner, which can be used for face recognition; Part of the contour points, including the contour points of the chin, that is, 68-point positioning. In face analysis tasks, it is essential to obtain facial feature points, such as face authentication and recognition, expression recognition, head pose estimation, 3D modeling of faces, and face beautification. [0003] Facial feature point acquisition methods can be roughly divided into three categories: optimization-based methods, regression-based methods, a...

AI Technical Summary

Problems solved by technology

[0009] The method based on regression has two main disadvantages: the effect of facial feature point acquisition depends on the initial shape, if the initial shape is far away from the target shape, the subsequent iterations of the cascade cannot correct the difference, and the regressed shape will fall into a local extreme. Small value; the shape index features used in the regression process are artificially designed or learned from shallow models, and it is difficult to accurately locate face images with special deflection poses

[0012] However, the inventors have found that in the prior art, the method for acquiring facial feature points based on convolutional neural networks adds multiple tasks (such as judging glasses, gender, smile and three-dimensional posture, etc.) while acquiring facial feature points, The convolutional neural network has a large number of layers and many parameters. Some methods integrate multiple networks, and some methods divide the face into multiple sub-regions, train a network for each region separately and then fuse them, making the amount of calculation Greatly increased, it takes up a lot of computing resources, and the training time is also greatly increased

[0013] For the above problems, no effective solution has been proposed

Images