A Human Behavior Recognition Method Based on Deep Recursive Hierarchical Conditional Random Field

Abstract

The present invention discloses a human body behavior recognition method based on depth recursive layered conditional random field. Firstly, the human body posture and the possible interaction with the behavior subject in the RGB-D video of the action scene captured by the RGB-D camera are respectively extracted. object information, and use these two types of information as the middle layer state of the deep recursive layered conditional random field, model the correlation between the current state in the predicted output target state layer and all the current predicted output state sets that have occurred, and construct a deep recursive classification Layer conditional random field model; secondly, the structured support vector machine classifier driven by BCFW optimization method is used to learn the discriminant classification model of human behavior sequence; finally, the human behavior to be tested is predicted according to the learned model parameters and the obtained discriminant model The category of the sequence. The invention has remarkable robustness to behaviors and actions, and improves the recognition accuracy of human behaviors and actions to a certain extent.

Description

technical field [0001] The invention relates to a human behavior recognition method, in particular to a human behavior recognition method based on Deep Recursive and Hierarchical Conditional Random Fields (DR-HCRFs), which belongs to the technical field of computer vision behavior recognition. Background technique [0002] Human behavior recognition plays an important role in computer vision, and has a wide range of applications in the fields of intelligent monitoring, human-computer interaction and sports video processing. [0003] In recent years, research on behavior recognition in indoor scenes mainly uses the method of probabilistic graphical model to classify and analyze human behavior. Common probabilistic graphical models are mainly divided into two structures: generative models and decision models. Common generative models include: Hidden Markov Model, Bayesian Networks (DBNs), and Semi-Markov Models. Generative models need to model the distribution and correlatio...

AI Technical Summary

Problems solved by technology

[0005] The above-mentioned existing behavior recognition methods based on probabilistic graphical models have not considered the internal representation of the target state and the high-order correlation between states, and still have the problem of low recognition accuracy

Images