Video-based human body action recognition method and device, medium and electronic device

Abstract

The invention relates to the field of computer vision, and particularly discloses a video-based human body action recognition method and device, a medium and an electronic device. The video-based human body action recognition method comprises the steps of obtaining multiple frames of multi-human body images in a video, wherein each frame of multi-human body image comprises multiple human body instances; generating a detection human body bounding box of a human body instance in each frame of multi-human body image; determining an image scene of each frame of multi-human body image, wherein one image scene corresponds to one or more types of actions; based on the detection human body bounding box and the image scene of the multi-frame multi-human body image, identifying the action of the human body instance in the multi-frame multi-human body image, and post-processing the action category and probability of the human body instance through the surrounding environment characteristics in the video frame of the human body instance in the video clip, so that the accuracy of the finally predicted action category of the human body instance is higher.

Description

technical field [0001] The present application relates to the field of computer vision, and in particular discloses a video-based human action recognition method, device, medium and electronic equipment. Background technique [0002] In many scenarios, it is necessary to recognize human actions in a video to determine the action type of the human action. For example, in a monitoring system, in order to perceive human intentions, it is necessary to recognize human actions in the video captured by the monitoring system; in a human-computer interaction system, in order to understand human behavior, it is necessary to recognize Human motion recognition in videos. [0003] In order to realize the recognition of human body movements in the video, the existing scheme is to extract some frames of pictures from the video to be processed, and then use the convolutional neural network to extract the features of the pictures, and then according to the position of the human body extract...

AI Technical Summary

Problems solved by technology

[0004] In the existing scheme, the features of some areas in the picture are directly extracted for action recognition, but the direct and simple extraction of features of some areas of the image may not be able to better reflect the action characteristics of the human body in the video frame scene, resulting in the effect of action recognition Not good, low accuracy

Images