Sit-up test counting method and sit-up test counting system based on Quick-OpenPose model

Abstract

The invention discloses a sit-up test counting method and system based on a Quick-OpenPose model, and the method comprises the steps: placing a camera on the side surface of a test region, and continuously photographing a test image of a testee at a certain frame rate; employing an improved Quick-OpenPose model for recognizing main parts of a human body, numbering and connecting the main parts ofthe human body, and selecting spine connecting lines and head and knee connecting lines; taking the angle between the spine connecting line and the ground as a preferred judgment condition, and when the spine connecting line cannot be recognized, selecting the length of the connecting line from the head to the knee as a standby judgment condition; setting three judgment conditions of lying up, sitting and lying down; when the three judgment conditions are all met, enabling the system to count once. Interference factors are reduced through visual recognition, equipment is simplified, and meanwhile, the invention further provides a corresponding counting system. Self-service sit-up testing and counting are achieved, the testing efficiency is high, the sit-up testing and counting device can be attached to an existing sit-up cushion, and real-time counting is achieved.

Description

technical field [0001] The invention relates to the technical fields of deep learning, machine vision and image processing, in particular to a method and system technology for sit-up test counting based on the Quick-OpenPose model. Background technique [0002] In the "National Physique Measurement Standards Manual and Standards (Adult Part)" issued by the Group Division of the State General Administration of Sport in 2003, sit-ups were listed as one of the test items for national physique. In modern life, the phenomenon of sub-health is serious, and sports such as fitness have become an upsurge, and sit-ups, as the basic movements in sports, have also risen accordingly. [0003] Under the increasingly higher requirements for the test accuracy of the items in the test and the need for unmanned testing, the existing sit-up test methods do not yet have a good test experience, and the test efficiency, accuracy, and degree of automation still need to be improved. . Existing si...

AI Technical Summary

Problems solved by technology

However, this method is not to detect the human body, but to identify the texture changes on the mat and then count the sit-ups. The quality of the sit-ups cannot be judged, and the recognition accuracy is greatly affected by the image texture interference.

The convolutional neural network test method is also limited by the recognition accuracy. However, after CAO Zhe et al. proposed the bottom-up OpenPose model in 2017, this method surpasses humans in r...

Images