A Novel Model for Intelligent Pull-Ups Test Based on Key Point Estimation of Human Body and Equipment

Table 5

Comparison of action classification conducted by PEPoseNet and MediaPipe in pull-ups test.

Metrics	Models	Standard actions	Nonstandard action
			Legs bent	Excessive swing	Nonstandard hands distance
Accuracy	PEPoseNet	0.980	0.990	0.993	0.994
	PEPoseNet-D	0.981	0.991	0.984	0.990
	MediaPipe	0.980	0.983	0.990	0.991
Precision	PEPoseNet	0.965	0.990	0.988	0.980
	PEPoseNet-D	0.974	0.969	0.969	0.978
	MediaPipe	0.974	0.967	0.977	0.968
Recall	PEPoseNet	0.982	0.972	0.978	0.982
	PEPoseNet-D	0.973	0.974	0.968	0.977
	MediaPipe	0.972	0.966	0.976	0.974
F1 score	PEPoseNet	0.974	0.981	0.983	0.981
	PEPoseNet-D	0.974	0.971	0.968	0.978
	MediaPipe	0.973	0.967	0.976	0.971

The values in bold represent this best data for this metrics.