Behavior recognition method and device and computer readable storage medium

Abstract

The invention relates to artificial intelligence, and provides a behavior identification method and device, and a storage medium, and the method comprises the steps: segmenting a video stream into animage frame sequence; detecting a human body contour in each frame of image, and marking each human body by using a first rectangular frame; calculating the distance between any two first rectangularframes in each frame of image; if the distance between the two first rectangular frames in a certain frame of image is smaller than a threshold value, surrounding the two first rectangular frames by atwo-person combined frame; searching a plurality of front and back frames of images, forming a two-person combination box by two persons the same as those in the two-person combination box, and forming a two-person combination box sequence by the frame of images and the two-person combination box in the plurality of front and back frames of images; and inputting the two-person combination box sequence into the neural network model for behavior identification. According to the method, a large amount of calculation amount of the neural network model caused by redundant backgrounds is avoided, the characteristics, valuable for behavior judgment, between two persons are reserved, and the accuracy of behavior recognition in a complex scene is improved.

Description

technical field [0001] The present invention relates to artificial intelligence, in particular to a behavior recognition method, device and computer-readable storage medium. Background technique [0002] At present, in the field of security protection, it is a common problem to judge the behavior of human body in the video. Detect events in the current area through the camera, for example: detect the active human body in the current area, and identify the behavior of the human body. In the traditional field of behavior recognition, the human body contour in the video frame is usually extracted, and the posture change of the human body contour is used to identify and judge which behavior category it belongs to. However, in a scene with a more complex background, or there are more When other people interfere, simply using the change of human body posture as a classification problem for behavior recognition will have a large misjudgment rate. [0003] In addition, for the beh...

AI Technical Summary

Problems solved by technology

In the traditional field of behavior recognition, the human body contour in the video frame is usually extracted, and the posture change of the human body contour is used to identify and judge which behavior category it belongs to. However, in a scene with a more complex background, or there are more When other people interfere, simply using the change of human body posture as a classification problem for behavior recognition will have a large misjudgment rate

[0003] In addition, for the behavior between two human bodies, the background between the two people has a very valuable basis for judging whether the behavior occurs, while the background outside the area between the two people is basically of little value for judging whether the behavior occurs. Moreover, for surveillance videos, usually the human body is small and the background is large, directly inputting the image of one frame in this video into the neural network model for calculation will inevitably cause a huge amount of calculation

Images