Neck rotation identification method and system

Abstract

The invention discloses a neck rotation identification method and system, and belongs to the three-dimensional identification technology. According to the method, a UV position map is designed, the UVposition map is a 2D image recording 3D coordinates of a complete face point cloud, and semantics are reserved at each UV position. Then, a simple encoder-decoder network with weighted loss is trained, which pays more effort on the feature region with better discrimination to return the UV position map from the single 2D facial image.

Description

technical field [0001] The invention relates to three-dimensional recognition technology, in particular to a neck rotation recognition method and system. Background technique [0002] For patients with nasopharyngeal carcinoma undergoing radiotherapy and chemotherapy, radiation exposure to the cervical spine may cause stiffness of the cervical spine joints, and they may even be unable to turn their heads. Therefore, patients need to perform neck training to prevent neck stiffness. There is currently no quantitative measure for the training effect. Therefore, it is planned to develop a neck joint Neck rotation training and recognition APP, that is, use the front camera of the Android phone to take pictures or use a small video to measure the angle of the horizontal rotation of the neck. If the steering angle is less than about 45°, it indicates that the cervical joints are beginning to harden and medical intervention is required. The key link in technical implementation is t...

AI Technical Summary

Problems solved by technology

The fundamental reason is that the facial landmarks are in a 2D low-dimensional space, and the solution of the facial pose needs to be upgraded to a 3D space, which results in endogenous limitations that cannot be broken through by this type of method.

Images