Method and terminal for identifying to-be-rescued person

Abstract

The invention discloses a method and a terminal for identifying a to-be-rescued person. The method comprises the steps of judging whether a first feature set of a to-be-identified person in a monitoring video can be successfully identified or not, wherein the identified first feature set meets a first preset condition within a preset time length; if yes, judging that the to-be-identified person isa to-be-rescued person; if not, continuing to obtain a second feature set of the to-be-identified person, judging whether a feature meeting a second preset condition exists in the second feature setor not; if yes, judging that the to-be-identified person is a to-be-rescued person, judging whether the to-be-identified person is a to-be-rescued person or not according to different feature sets, comprehensively judging the to-be-identified person, avoiding omission of the to-be-rescued person, and improving comprehensiveness and accuracy of judging whether the to-be-rescued person is a to-be-rescued person or not.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a method and terminal for identifying persons to be rescued. Background technique [0002] In daily life, there are often incidents where the elderly fall or encounter unexpected situations. If no one finds out at this time, it will bring certain safety hazards. At present, there is an automatic fall detection method. By extracting the feature points of the target person in the video, the fall judgment is made from multiple directions of the target person. When the judgment result of any direction shows that the target person is in a falling state, the continuous certain frame Whether the target person keeps falling within a few days, and if so, confirm that the target person has fallen. [0003] Specifically, by determining the bone center point, shoulder joint center point, and waist center point of the human body in each frame of the video, the angle formed by...

AI Technical Summary

Problems solved by technology

[0003] Specifically, by determining the bone center point, shoulder joint center point, and waist center point of the human body in each frame of the video, the angle formed by the upper body torso center line of the human body and the gravity line and the angle between the center line and the preset time length can be calculated. The rate of change and the rate of change of the distance between the waist center point and the reference plane or reference point, when the rate of change of the included angle is greater than the first preset value and the rate of change of the distance is greater than the second preset value, it is determined that the human body has fallen , or starting from the height, connect the bone points such as the head, neck, spine, hips and knees, and detect whether the connection line is parallel to the ground and whether the height of the spine is less than a given threshold and exceeds the preset time to detect falls. , send a signal for help after a fall is detected, but this method only sends a signal after judging the position of the human torso, which is prone to misjudgment, is not accurate enough, and is not comprehensive enough

Images