Table 2.

The APs of the two baseline detectors, namely CHP-YOLOF and CHP-YOLOX, are evaluated on the CHP dataset. It is important to note that $A{P}_c$, $A{P}_p$, and $A{P}_{cp}$ are the accuracy metrics employed to predict human, common human posture, and combinations of both.

Model	{AP${}_{\mathit{c}}$, AP${}_{\mathit{p}}$, AP${}_{\mathit{cp}}$}@0.5	{AP${}_{\mathit{c}}$75, AP${}_{\mathit{p}}$75, AP${}_{\mathit{cp}}$}@0.75	{mAP${}_{\mathit{c}}$, mAP${}_{\mathit{p}}$, mAP${}_{\mathit{cp}}$}
CHP-YOLOF	(0.818, 0.652, 0.708)	(0.720, 0.579, 0.632)	(0.608, 0.488, 0.531)
CHP-YOLOX	(0.828, 0.696, 0.727)	(0.692, 0.579, 0.612)	(0.588, 0.490, 0.516)