Illustration of the balance boards, which could freely move in the frontal plane, varying in height of the surface of the board above the point of contact with the floor (BB1; 15 cm, BB2; 17 cm and BB3; 19 cm).

Contexts in source publication

Context 1

... participants performed bipedal upright standing on a rigid surface and on three balance boards varying in height of the surface of the board above the point of contact with the floor (BB1; 15 cm, BB2; 17 cm and BB3; 19 cm). The balance board was a 48 cm by 48 cm wooden board mounted on a section of a cylinder with a 24 cm radius that could freely move in the frontal plane ( Figure 1). The four conditions, were repeated three times in random order, with each trial lasting 16 seconds. ...

Context 2

... Mathis et al., 2018;Nath et al., 2019). A full-body human model consisting of 48 markers was manually labelled in the extracted frames ( Figure A.1.). These data were used to train a deep neural network (resnet_50) in DeepLabCut for each camera with a training fraction of 0.95 and the following learning rates; lr=0.005 for 10k, then lr=0.02 for 430k, lr=0.002 for 730k and lr=0.001 for 1.3M iterations (Table A.1.). ...

Context 3

... data were used to train a deep neural network (resnet_50) in DeepLabCut for each camera with a training fraction of 0.95 and the following learning rates; lr=0.005 for 10k, then lr=0.02 for 430k, lr=0.002 for 730k and lr=0.001 for 1.3M iterations (Table A.1.). The trained networks were used to track 2D coordinates from the videos of the eight cameras ( Figure A.1.). To acquire 3D coordinates, Anipose was used to triangulate with optimization across camera views using the calibration data as extracted from the standard Simi calibration (Figure A.1.). ...

Context 4

... trained networks were used to track 2D coordinates from the videos of the eight cameras ( Figure A.1.). To acquire 3D coordinates, Anipose was used to triangulate with optimization across camera views using the calibration data as extracted from the standard Simi calibration (Figure A.1.). A Simi motion wand and L-frame calibration was performed to calibrate the cameras and the calibration file was converted into the "openCV" format which Anipose uses, to be able to triangulate across the camera views (https://osf.io/7hn8z/). ...