Human skeleton action prediction method based on multi-task non-autoregressive decoding

Abstract

The invention discloses a human skeleton motion prediction method based on multi-task non-autoregressive decoding, which is used for solving the problem of motion prediction of a human 3D skeleton. The method specifically comprises the following steps of obtaining a human body 3D skeleton key point data set for training, and defining an algorithm target; establishing a graph convolution encoder, and performing feature learning on the input human body 3D skeleton to obtain features of the input skeleton; establishing a classifier, and performing behavior recognition on the input human body 3D skeleton input; establishing a non-autoregressive decoder, and predicting a human body 3D skeleton at a future moment; performing behavior recognition on the predicted human body 3D skeleton by using ashared graph convolution encoder and a classifier; and using the joint learning framework to carry out human body action prediction at a future moment. The method is used for human body action prediction analysis in a real video, and has good effect and robustness for various complex conditions.

Description

technical field [0001] The invention belongs to the field of computer vision, in particular to an action prediction method of a human skeleton based on multi-task non-autoregressive decoding. Background technique [0002] The action prediction problem based on the human skeleton is defined as the following problem: in a set of human skeleton key point sequences containing multiple frames, predict the human key point sequence at a future moment. Human skeleton key points are often used as auxiliary information for some high-level visual tasks, such as human-computer interaction, abnormal behavior detection, etc. The key factors of human skeleton action prediction include modeling the temporal structure. Traditional methods often use autoregressive decoding, that is, the prediction result of the next frame is completely dependent on the prediction result of the previous frame, and this recursive method is used to predict the future time, without considering the prediction err...

AI Technical Summary

Problems solved by technology

Traditional methods often use autoregressive decoding, that is, the prediction result of the next frame is completely dependent on the prediction result of the previous frame, and this recursive method is used to predict the future time, without considering the prediction error will also follow the non-autoregressive The decoding process propagates the fact that

Images