Method for estimating three-dimensional human body posture from video

Abstract

The invention discloses a method for estimating a three-dimensional human body posture from a video, and the method comprises the steps of obtaining a continuous video frame sequence, extracting a single-frame human body posture based on a first neural network model, and obtaining an initialized two-dimensional human body joint of each frame; extracting three-dimensional human body joint estimation of N frames of initialized two-dimensional human body joints based on a second neural network model to obtain initialized three-dimensional human body joints; acquiring M continuous frames of initialized two-dimensional human body joints, and performing three-dimensional joint filtering on the M frames of initialized two-dimensional human body joints based on a third neural network model to obtain a three-dimensional human body joint filter about the initialized two-dimensional human body joints; and denoising the initialized three-dimensional human body joint by using a three-dimensional human body joint filter to obtain a denoised three-dimensional human body joint. The BVH-format human body posture can be directly estimated through the video, the BVH-format human body posture has a good visual effect, and the final human body posture estimation result has a better animation role driving effect.

Description

technical field [0001] The invention relates to the technical field of three-dimensional human body estimation, in particular to a method for estimating a three-dimensional human body pose from a video. Background technique [0002] 3D human body pose estimation is a technology that can directly obtain 3D human body poses from videos. It is widely used in video evaluation and human body pose analysis in videos. It is very important for judging the motion state of the human body. Analytics is a huge boost. [0003] In the 3D pose estimation network, the data in the 3D coordinate format is almost always used as the estimation result, but once such a 3D coordinate format becomes the ground truth data, the 3D pose obtained by the network exists in the 3D coordinate format , but the three-dimensional coordinate format data is not suitable for driving animation characters due to the lack of rigid body constraints. In this way, when faced with the driving requirements of animation...

AI Technical Summary

Problems solved by technology

[0003] In the 3D pose estimation network, the data in the 3D coordinate format is almost always used as the estimation result, but once such a 3D coordinate format becomes the ground truth data, the 3D pose obtained by the network exists in the 3D coordinate format , but the three-dimensional coordinate format data is not suitable for driving animation characters due to the lack of rigid body constraints. In this way, when faced with the driving requirements of animation characters, none of the existing network technical solutions can adapt

Images