Unlock instant, AI-driven research and patent intelligence for your innovation.

Human interaction behavior recognition method, system and device based on multi-feature fusion

A technology of multi-feature fusion and recognition method, which is applied in the field of system and its device, and the recognition method of two-person interactive behavior, which can solve the problems of recognition error, loss of interactive information, and loss of information, etc.

Active Publication Date: 2022-03-22
NORTH CHINA UNIVERSITY OF TECHNOLOGY
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in the process of feature extraction, if a single person’s features are used to predict the overall interaction, the result will be affected by the irrelevant information of a single person’s unique action in the action process. If the features of two people are extracted separately and then Combining will lose the information of the interaction between the two people, but this point is not considered in the existing technology, which leads to the loss of many important interaction information, recognition errors and other problems

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Human interaction behavior recognition method, system and device based on multi-feature fusion
  • Human interaction behavior recognition method, system and device based on multi-feature fusion
  • Human interaction behavior recognition method, system and device based on multi-feature fusion

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0074] In a specific embodiment, taking the recognition of two-person interaction behavior as an example, the technical solution of the present invention can be realized through the following implementation modes, that is, through a single-person feature extraction module, a feature fusion module, a human body interaction information extraction module, and a human body Each method step corresponding to the interactive behavior classification recognition module is realized:

[0075] 1. Single-person feature extraction module

[0076] This module processes single-person features, regards interactive behavior as two independent individuals, and extracts local spatiotemporal features and global optical flow features for each individual.

[0077] Step 1: Extract optical flow and describe it

[0078] Calculate motion region optical flow. In a preferred implementation, the optical flow estimation using the inter-frame difference can be specifically performed in the following way: a...

Embodiment 2

[0140] In another specific embodiment, the technical solution of the present invention can also be realized by a human interaction behavior recognition system based on multi-feature fusion, and the system can execute the specific method described in Embodiment 1.

[0141] Specifically, preferably, the system may include:

[0142] The single-person feature extraction module extracts local spatio-temporal features and global optical flow features for multiple independent individuals in the interactive video;

[0143] A feature fusion module, configured to perform feature fusion of the local spatiotemporal features and the global optical flow features, obtain the correlation between the local spatiotemporal features and the global optical flow features, and then obtain the single Personal characteristics information;

[0144] The interactive information extraction module is used to pass the single-person feature information of each independent individual obtained by the feature ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The present invention provides a method, system and device for human interaction behavior recognition based on multi-feature fusion. The method includes: extracting local spatio-temporal features and global optical flow features from multiple independent individuals in the interactive video; The spatio-temporal feature is fused with the global optical flow feature to obtain the correlation between the local spatio-temporal feature and the global optical flow feature, and then obtain the single-person feature information of each independent individual; through the parallel long-term short-term memory network, obtain Mixed feature information; based on mixed feature information, classify and obtain interactive behavior recognition results. The technical scheme of the invention fully retains the behavior information and interaction information in the interaction, and greatly improves the accuracy of interaction behavior recognition.

Description

technical field [0001] The invention belongs to the field of computer vision and can be used for the analysis and research of human interaction behavior, and in particular relates to a recognition method, system and device for two-person interaction behavior based on multi-feature fusion. Background technique [0002] Human interaction behavior recognition is one of the research hotspots in the field of computer vision, which not only has important theoretical significance, but also has broad application prospects. The intelligent video surveillance system using human interaction technology is an important part of the urban security system, and will be widely used in places such as banks, stations, schools, and military areas where people are dense, difficult to manage, or sensitive to security requirements. ; In the huge video database, human interaction behavior recognition can be automatically retrieved according to the predefined behavior patterns, making it very conveni...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06V40/10G06V10/764G06V10/80G06K9/62
CPCG06V40/10G06F18/241G06F18/254
Inventor 叶青杨航张永梅
Owner NORTH CHINA UNIVERSITY OF TECHNOLOGY