Image data calibration training method and device, storage medium and server

Abstract

Embodiments of the invention disclose an image data calibration training method and device, a storage medium and a server. The method comprises the following steps of obtaining a corresponding disturbed image data set after disturbance processing of image data, and reference calibration coordinates corresponding to the set; based on a CNN, obtaining target calibration coordinates of the disturbed image data set; taking the reference calibration coordinates and the target calibration coordinates as input parameters of a forward propagation function in an image data calibration training process; obtaining gradient information of the forward propagation function, and obtaining an output change quantity of the forward propagation function according to the gradient information and a learning rate of the CNN; according to the output change quantity, obtaining a coordinate change quantity of the target calibration coordinates in a reverse propagation function in a training process; and based on the coordinate change quantity, adjusting network parameters of each layer in the CNN to minimize the output change quantity of the forward propagation function obtained by calculation of the CNN subjected to the network parameter adjustment. By adopting the method and the device, the accuracy of calibrating a human face in a picture by a trained model can be improved.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to an image data calibration training method and equipment, a storage medium, and a server. Background technique [0002] In the era of the Internet with the rapid development of Internet technology and the increasingly mature artificial intelligence technology, face positioning, an image processing technology, has become one of the hot research technologies at present. Among them, face positioning is a technology that calculates the position of each feature point of the face on the picture through an image processing algorithm. The feature point can be a point with a significant price comparison on the face, such as the corners of the eyes, eyebrows, mouth, nose tip, and chin. In the prior art, a deep learning algorithm (for example, convolutional neural network CNN) is currently the most commonly used face localization algorithm. Specifically, the training model obtained by trai...

AI Technical Summary

Problems solved by technology

In the prior art, a deep learning algorithm (for example, convolutional neural network CNN) is currently the most commonly used face localization algorithm. Specifically, the training model obtained by training the images in the image database can be used to identify the features of each feature point on the face. However, due to the errors or errors in the calibration points of some images in the image library used for deep learning training, it affects the accuracy of the trained model to calibrate the facial feature points in the image.

Images