Unlock instant, AI-driven research and patent intelligence for your innovation.

Head posture estimation method, computer equipment and storage medium

A technology for head pose and model estimation, applied in the field of image processing, can solve problems such as low accuracy, and achieve the effect of large amount of data, rich data diversity, and simple and efficient acquisition.

Active Publication Date: 2022-04-15
小视科技(江苏)股份有限公司
View PDF3 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The more mainstream method is to pre-set a set of default three-dimensional coordinates of face key points, use the PnP algorithm to calculate the approximate transformation matrix from the default three-dimensional key point projection to the detected key point, and then extract the rotation matrix. Affected by camera parameters and face shape, so the accuracy is low

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Head posture estimation method, computer equipment and storage medium
  • Head posture estimation method, computer equipment and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0030] The present invention will be described in detail below in conjunction with the accompanying drawings.

[0031] This embodiment provides a head pose estimation method, such as figure 1 shown, including the following steps:

[0032] 1. Use the 3DMM model to generate multi-pose 3D face data. Face 3DMM (3D Morphable Models, three-dimensional deformable model) is a parametric model of the face, which can generate a 3D face shape with a specific posture by adjusting identity parameters, expression parameters, and rotation angles. Common face 3DMM models include BFM (Basel Face Model, Basel Face Model), LSFM (Large Scale Facial Model, large-scale face model), etc.

[0033] Select a 3DMM model close to the target scene, randomly adjust the identity coefficient, expression coefficient, and rotation angle to generate N (N is a positive integer greater than 100) 3D faces , and record the head pose corresponding to each 3D face ,in Refers to the nth (n is 0 to N-1) 3D face...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a head posture estimation method, computer equipment and a storage medium, and the method comprises the steps: generating N 3D human faces through employing a three-dimensional deformation model, and recording the head posture corresponding to each 3D human face; acquiring key points of the 3D face; constructing a training data set by adopting the N head postures and the corresponding key points, and training a head posture estimation model; and carrying out key point detection on a to-be-recognized face image, and inputting the detected key points of the to-be-recognized face image into the trained head posture estimation model to obtain a head posture result. The training data set is simple and efficient to obtain, the generated data is rich in diversity and large in data volume, the scene influence in the image is eliminated by predicting the head posture from the key point coordinates, and the stability is high.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a head pose estimation method, computer equipment and a storage medium. Background technique [0002] Head pose estimation refers to predicting the three rotation angles (pitch angle, yaw angle, and barrel roll angle) of a person's head based on the face image. [0003] The traditional method of head pose estimation method, one is to predict the head pose directly from the face image, which needs to use 3D sensors to collect the head poses of different people and different angles, and then train the machine learning model with this, the data acquisition cost is high, Moreover, the data scenarios (background, lighting, identity) collected in the laboratory environment are single, and the data diversity is insufficient. [0004] There is another method that divides pose estimation into two steps: face key point detection and pose prediction from key points. Since ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06V40/16G06V20/64G06N3/08
CPCG06N3/08
Inventor 杨帆郝强潘鑫淼胡建国
Owner 小视科技(江苏)股份有限公司