Fig 1.
Twenty five joints of the virtual skeleton characteristic to the second generation of Kinect.
The 17 joints that were analysed in this study (see Methods) are marked as white, as opposed to the skipped eight joints, which are marked as black.
Fig 2.
A young athlete training with the Kinect-based system.
Graphical information displayed on the screen provides feedback (see text). The parent of the young individual depicted in the figure has given written informed consent (as outlined in the PLOS consent form) to publish these case details.
Fig 3.
The real-time (left) and post-hoc (right) feedback presented visually to the athlete during Kinect aided training.
In the real-time feedback (left), the master skeleton is displayed in yellow, whereas the skeleton of the practicing athlete is simultaneously displayed in green. The z-error (along the optical axis of the Kinect) of a joint position is colour coded in blue/red when the joint is located further/closer from Kinect than expected with respect to the master performance. In the post-hoc feedback (right), the size of the x-, y-, and z-error is graphically presented in red, green, and blue, respectively, for each of the joints.
Fig 4.
Panels in the left column show median error across athletes in a group and trials in a training session, in metres, with respect to the master performance for 17 body joints labelled along the horizontal axis. Consecutive rows show four types of comparisons: Aided vs Controls in the 1st and 5th session, the 1st vs 5th session for Aided and Controls; see legends. Solid triangles denote statistically significant differences with respect to the values represented by open circles, with grey and black triangles for p < 0.05 and p < 0.01, respectively. The right-column panels show the median of differences, marked with a black dot, with the 95% confidence interval bounded by the two triangles. The black bar limits correspond to the 25th and 75th percentiles, respectively. The whiskers extend to the most extreme data points not considered outliers, and the outliers are plotted individually as circles. The green stars denote the effect size defined as , where z is the value of the z-statistic of the Wilcoxon rank sum test or the Wilcoxon signed rank test and N is the sample size.
Fig 5.
Like in Fig 4 but for osoto-gari.
Fig 6.
Like in Fig 4 but for ouchi-gari.
Fig 7.
p-values (a1–d1) and z-statistics (a2–d2) for the comparisons depicted in Fig 4 (in red), Fig 5 (in green), and Fig 6 (in blue).
Values are emphasized with bold font when p < 0.05.
Fig 8.
The difference of Controls’ performance in their 1st and 5th session vs the corresponding difference of Aided, for de-ashi-barai.
For better readability, scales are different than those in Figs 4–6.
Fig 9.
Like in Fig 8 but for osoto-gari.
Fig 10.
Like in Fig 8 but for ouchi-gari.
Fig 11.
p-values (left) and z-statistics (right) for the comparisons depicted in Fig 8 (in red), Fig 9 (in green), and Fig 10 (in blue).
Values are emphasized with bold font when p < 0.05.