Action detection method and device and storage medium

Abstract

The invention provides an action detection method and device and a storage medium, and relates to the technical field of image processing. The motion detection method comprises the following steps: acquiring the position of a joint point of a detected person according to a video image, and generating a skeleton diagram of the detected person; obtaining a skeleton sequence diagram according to the skeleton diagrams of a predetermined number of tested persons within a predetermined time length; according to the skeleton sequence diagram, whether the detected person executes a target action or not is determined through a neural network model, a judgment result is output, and the neural network model is generated through training of the skeleton sequence diagram determined to be under the condition that the person executes the target action and the skeleton sequence diagram determined to be under the condition that the person does not execute the target action. Through the method, the dynamic state of the skeleton in the human body movement process can be extracted from the image, the human body movement is determined through the neural network model based on the change of the skeleton position, the posture change implicit information of a person under the movement is fully utilized, and the accuracy of movement determination is improved.

Description

technical field [0001] The present disclosure relates to the technical field of image processing, in particular to an action detection method, device and storage medium. Background technique [0002] At present, traditional fall detection methods based on wearable devices and audio sensors have shortcomings such as high intrusion, low precision, poor robustness, and susceptibility to environmental influences. With the development of computer vision and artificial intelligence technology, vision-based fall detection methods can overcome the shortcomings of traditional wearable devices and sensor methods and are more acceptable, and can be widely applied to IoT environments such as smart homes and smart homes. Contents of the invention [0003] One object of the present disclosure is to improve the accuracy of image-based motion detection. [0004] According to an aspect of some embodiments of the present disclosure, an action detection method is proposed, including: acquir...

AI Technical Summary

Problems solved by technology

[0002] At present, traditional fall detection methods based on wearable devices and audio sensors have shortcomings such as high intrusion, low precision, poor robustness, and susceptibility to environmental influences.

Images