Human Pose Detection Method Based on Monocular RGB Image

Abstract

The invention discloses a human body posture detection method based on a monocular RGB image, which includes S1, using a monocular RGB camera to collect video data, and using a frame filtering method to extract image frames in the video data, and then preprocessing the video data ; S2. Input the preprocessed image frames into the human body key point detection model to extract human body key points, and calculate the human body characteristic parameters of the image frame based on the human body key points, and mark the current human body characteristic parameters as F t ; S3, judgment F t Whether the complete human body structure information is detected for the first time, and if so, the initial comparison reference value is F t , otherwise adopt the method of refreshing the comparison reference quantity to update the comparison reference quantity; S4, according to the human body characteristic information parameter and the comparison reference quantity, use the threshold judgment method to detect the human body posture; S5, when the posture of the human body is falling, send an alarm.

Description

technical field [0001] The invention relates to fall detection in the field of personnel safety monitoring, in particular to a human body posture detection method based on a monocular RGB image. Background technique [0002] As China's population aging problem continues to aggravate, the elderly's pension safety monitoring issue has received more and more social attention. At present, the most important problem facing the field of personnel safety monitoring is fall detection. [0003] There are three main types of core methods for human fall detection: one is to collect the height of the center of mass of the human body through wearable sensors for real-time monitoring, and transmit the collected real-time data back for fall detection. The biggest disadvantage of this method is that wearable devices Easy to damage, low physical comfort; the second is to detect falls by deploying environmental sensors in the room, collecting and analyzing the sound, floor vibration and opti...

AI Technical Summary

Problems solved by technology

[0005] In view of the above-mentioned deficiencies in the prior art, the human body posture detection method based on the monocular RGB image provided by the present invention solves the problem that the fall detection accuracy rate of the fall detection method in the prior art is not high

Images