Text this: Markerless versus marker-based 3D human pose estimation for strength and conditioning exercise identification.