Human body attitude tracking analysis-based indoor dangerous condition warning method

Abstract

The invention relates to a human body attitude tracking analysis-based indoor dangerous condition warning method. According to the method, an artificial intelligence network-based human body recognition technology is applied to household appliances such as air conditioners, and therefore, the original monotonous functions of the household appliances can be improved, functions for judging and identifying dangerous behaviors of children which may cause harm, such as moving the household appliances, climbing to high positions and touching sockets, can be increased, and warning can be triggered timely when the dangerous behaviors are detected; functions for judging and identifying the accidental tumble or specific help-seek gestures of old people at home can be increased, and warning can be triggered timely; and whether dangerous conditions such as a fire, occur in a house can be judged, when the dangerous conditions are detected, warning can be triggered timely.

Description

technical field [0001] The invention relates to a method for warning indoor dangerous situations based on human body posture tracking analysis. Background technique [0002] Motion Capture is a measurement technique that records the movement of an object and simulates it into a digital model. Motion capture involves various calculation methods such as measurement, physical positioning, and spatial positioning, as well as the intercommunication and processing between data and computers. By setting the tracker on the key part of the moving object or obtaining the position of the target point by other means, and then get the data of the three-dimensional space coordinates after computer processing, and apply the data in animation production, gait analysis, biomechanics, man-machine Engineering and other fields. [0003] Human body recognition refers specifically to computer technology that uses analysis and comparison of human visual feature information to distinguish people ...

AI Technical Summary

Problems solved by technology

These problems are difficult to solve by traditional rule-based programming

[0007] In the current human motion capture and posture analysis system, the background difference method is usually used for foreground detection, and the Gaussian model is often used for background modeling. This method is computationally intensive, slow, and easily affected by shadows. The detected target image has noise, and Targets that pause motion are absorbed as part of the background, resulting in holes inside the target

Moreover, the extraction of human joint points usually adopts optical marking or manual calibration, which has a large human error and a low degree of automation. Moreover, the current automatic extraction method of human joint points usually adopts the method of curve fitting, which has a large amount of calculation and low accuracy. universal applicability

[0008] In addition, there is no solution for human body posture analysis and tracking on air conditioners, and it is impossible to detect abnormal human movements, such as elderly people falling down, children falling down, or moving heavy appliances and other behaviors that may cause danger. alarm