Method for constructing human body posture reference image library

Abstract

The invention discloses a method for constructing a human body posture reference image library. According to the method, different human body posture movements are sequentially constructed into an induction video game; a player executes the corresponding movements according to the induction video game; image acquisition equipment is utilized to acquire a posture image of the player at a preset time point or a preset event position; the posture images of different players at the same time point or the same event position are classified into the same type; a plurality of posture images in the same type are subjected to feature extraction and clustering analysis; the posture images with low confidence coefficients are deleted; the left posture images with high confidence coefficients are correlated with the human body posture movements at the corresponding time points or the corresponding event positions in the induction video and are labeled; the human body posture reference image library is constructed. The method has the obvious effects that the operation of collecting a great number of human body posture pictures and then carrying out manual screening and calibration does not need to be carried out; a great amount of manpower and material resources are saved; meanwhile, a complex human body posture identification algorithm does not need to be used; the efficiency is improved.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a method for constructing a reference image database of human body posture. Background technique [0002] In the field of artificial intelligence, human body posture recognition has important theoretical research and engineering practice value. A human pose benchmark image library with rich samples and high quality is a basic tool for human pose recognition research, which can be used for algorithm training, testing and verification, so as to promote the development of human pose recognition technology. [0003] At present, there are two main methods to obtain the benchmark database of human body poses: one is to select a large number of pictures and obtain the pose database through manual calibration; the other is to select a large number of pictures and classify the pictures through some human body pose recognition algorithms to obtain the body Pose benchmark d...

AI Technical Summary

Problems solved by technology

The limitation of the above-mentioned method 1 is: firstly, a large number of pictures need to be collected and selected, which requires a certain amount of manpower and material resources

For these pictures, it is also necessary to determine their category through human posture judgment, which will inevitably consume more time and resources

Moreover, human judgment of posture will also receive interference from some subjective factors, which will also affect the quality of the final human posture benchmark database

[0004] The limitation of the above method 2 is the same as that of method 1, and the selection of a large number of pictures will consume a lot of resources

At the same time, most of the human body posture recognition algorithms in the field of artificial intelligence still have many defects: the recognition of human body movement postures is mainly concentrated on simple human movement postures, such as some standard postures of people and some simple movement behaviors, such as Simple and standardized actions such as walking, running, squatting, and standing are difficult to recognize some complex postures; the use of machine learning tools to identify human action behavior is still in its infancy, and constraints need to be added to the human motion model during the recognition process to reduce the Ambiguity, and these constraints are usually inconsistent with the general reality

This method is similar to the method of obtaining the pose database through manual calibration. Although it can provide data support for users to find the target video scene clips conveniently and quickly, the time anchor mark and subtitle annotation of the video scene in the video library It will consume a lot of time and manpower, and may also cause some subjective errors, which will affect the quality of the final generated database.

Images