Elevator car door pushing behavior recognition method based on machine vision

Abstract

The invention provides an elevator car door pushing behavior recognition method based on machine vision. The elevator car door pushing behavior recognition method comprises the steps of: 1, collecting video samples, containing a plurality of scenes in an elevator car, of a network camera; 2, extracting image frames from the video samples at intervals to make a target detection data set; 3, performing YOLOV5 network training on the target detection data set, extracting a personnel target in the elevator car, and storing a target frame identification area of the personnel target as a picture; 4, labeling 18 key points of a human body pose according to the stored picture, and making a human body pose key point data set; 5, conducting Openpose network training on the human body pose key point data set, and extracting 18 key points of the pose of each person target in the elevator car; and 6, judging whether a door pushing behavior exists or not by utilizing the extracted 18 key points of the pose of each person target in the elevator car.

Description

technical field [0001] The invention relates to a machine vision-based recognition method for door-snapping behavior of an elevator car. Background technique [0002] With the increase of high-rise buildings, the safety of elevator cars has been paid more and more attention by people. During the operation of the elevator, especially when it stops running due to some faults, some passengers often fail to take the correct handling method, and escape by pulling the elevator door, which brings greater danger or even life-threatening. In order to avoid the serious harm caused by door-slamming, the elevator operation and management department should deal with the failure of the elevator as soon as possible on the one hand, and on the other hand, it needs to detect the behavior of passengers in the elevator car, and stop the door-slamming behavior in time to avoid serious consequences. [0003] Most of the current elevator car monitoring relies on manual monitoring to detect car a...

AI Technical Summary

Problems solved by technology

There are also many misjudgments simply relying on the way that there are passengers in the door area when the elevator door is cracked. For example, the elevator is malfunctioning, the door is not closed tightly, or there are many people in the car. Passengers are always waiting near the door

Images