Human motion capture and virtual animation generation method based on deep learning

Abstract

The invention discloses a human motion capture and virtual animation generation method based on deep learning. The human motion capture and virtual animation generation method comprises the followingsteps: A, collecting actors' motions and converting the actors' motions into video stream signals for input; b, pre-training a posture detection network; c, extracting a human body posture sequence; d, pre-training the posture condition generative adversarial network; and E, inputting the posture sequence into the posture condition generative adversarial network, and outputting an animation videosynchronous with the human body action. According to the invention, the animation production cost is effectively reduced; the operability of a common user is improved; meanwhile, the output efficiencyof animations is also improved, and the method can be used as a media creation tool for special effect demonstration, real-time generation of Demo and rapid production of animations and movies, and can also be used as an interactive filter in short video application, assistance of virtual reality somatosensory games and the like, so that the large-range landing commercial and popularization of the action capture technology are facilitated.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a deep learning-based human motion capture and virtual animation generation method. Background technique [0002] Most of the current digital film and animation production processes use motion capture (Motion capture) technology. The traditional method records and captures real motion information through sensors or markers worn on the actor's body, and then restores and renders these motions to the corresponding virtual reality. Image body, get the corresponding virtual animation effect. [0003] However, this technology also has corresponding disadvantages: on the one hand, professional motion capture systems are expensive and not suitable for large-scale commercial use, and there are strict requirements on the lighting and reflection of the performance venue, and the device calibration process is also relatively cumbersome. This undoubtedly restricts its application in somat...

AI Technical Summary

Problems solved by technology

[0003] However, this technology still has corresponding shortcomings: on the one hand, professional motion capture systems are expensive and not suitable for large-scale commercial use, and there are strict requirements on the lighting and reflection conditions of the performance venue, and the device calibration process is also relatively cumbersome.

This undoubtedly restricts its application in somatosensory interactive games, and also hinders the experience of ordinary users

In addition, although it can capture real-time motion, manual intervention is required for post-processing (identification, tracking, and reconstruction of marker points) to apply these data to the animation character model, and the workload of animation rendering in the later stage is relatively large.

Images