A method and device for constructing human body posture data samples

Abstract

The invention discloses a method and device for constructing human body posture data samples, including constructing a skeleton model according to preset parameters; constructing several motion models between joints in the skeleton model, each motion model corresponding to a preset human motion , and the motion model includes the relative rotation relationship of each joint under the corresponding human motion; render the skeleton animation according to the motion model, and obtain animation images under various preset human motions; adjust the observation field of view, and repeat the operation of this step to obtain the skeleton The animation images of the model in each observation field of view are saved; adjust the parameters of the skeleton model, rebuild the skeleton parameters, until the animation images corresponding to the skeleton models under all preset parameters are obtained, and then extract the real value data according to the obtained animation images. The value data and animation images are human body pose data samples. The invention can generate a large number of human body posture data samples through the established skeleton model, saves manpower and has high construction efficiency.

Description

technical field [0001] The invention relates to the technical field of machine learning, in particular to a method and device for constructing human body posture data samples. Background technique [0002] Machine learning is an important branch of artificial intelligence. Machine learning is to let the machine learn the internal laws between samples through known samples. [0003] In the application of machine learning, there are more and more applications of human body posture data samples for human behavior recognition, human skeleton line recognition and other operations. Since commonly used machine learning requires at least hundreds of thousands of sample sets, if so many samples are only manually constructed, it will be time-consuming, labor-intensive, and inefficient. [0004] Therefore, how to provide a method and device for constructing human body pose data samples that improves sample construction efficiency is a problem that those skilled in the art need to solv...

AI Technical Summary

Problems solved by technology

Since commonly used machine learning requires at least hundreds of thousands of sample sets, if so many samples are only manually constructed, it will be time-consuming, labor-intensive, and inefficient

Images