A hybrid motion capture system based on deep learning

Abstract

The embodiment of the invention discloses a hybrid motion capture system based on deep learning. The system comprises: an inertial motion capture module, which is used for acquiring inertial attitudeinformation of an object; an optical motion capture module, configured to acquire optical attitude information of the object; a communication module, configured to send the inertial attitude information and the optical attitude information to the analysis module; and an analysis module, which is used for fusing the inertial attitude information and the optical attitude information according to thepreset hybrid model to obtain the optimal motion estimation of the object. In this way, through the characteristics that inertial motion capture is unobstructed and optical motion capture data is accurate, and through deep learning training, a hybrid model which integrates the inertial motion capture and optical motion capture characteristics is obtained, ensuring the accuracy of the final data.

Description

Technical field [0001] The present invention relates to the field of computer technology, in particular to a hybrid motion capture system based on deep learning. Background technique [0002] Traditional computer vision 3D reconstruction has the characteristics of high accuracy, high cost, and high delay, while the motion capture technology used in virtual reality has the characteristics of low cost, low delay and high accuracy. At present, the main domestic motion capture technology can be divided into inertial motion capture technology, optical motion capture technology and other motion capture methods. Inertial motion capture technology uses inertial sensors to collect the motion information of each joint of the moving object. The capture accuracy is not high, but the real-time performance is good, and it can collect occluded motion; optical motion capture uses one or more cameras to capture the moving object at the same receiving point. Parallax calculates its spatial positi...

AI Technical Summary

Problems solved by technology

Inertial motion capture technology uses inertial sensors to collect motion information of each joint of a moving object. The capture accuracy is not high, but it has good real-time performance and can collect occluded motion; optical motion capture uses one or more cameras to capture motion information generated by the same receiving point of a moving object. Parallax calculates its spatial position and rotation information, and the capture accuracy is high, but it is easily affected by motion occlusion

Images