Based on my research activities, I developed some demos to show examples of potential applications offered by human motion analysis. The following applications are currently “beta” to illustrate my research.

Visualization on Sphere

The aim of this application is to illustrate my research on human action recognition. It allows to visualize actions lying on the high-dimensional Riemannian shape. From the MSR Action 3D dataset, I used one representant for each action class. To reduce the dimensionnality and observe data on a sphere, I used Multidimensional Scaling (MDS) with the elastic distances between representative actions. From these existing actions, one can generate new actions represented by a weighted Riemannian center of mass among a set of existing actions.
More details can be found in the paper: “3-D Human Action Recognition by Shape Analysis of Motion Trajectories on Riemannian Manifold” 

Legend for the demo
 Existing actions
Highlighted actions
(when cursor is over)
Clicked point
Neighboring actions

Gesture Learning

This demo illustrates an example of gesture assesment. For instance, it could be useful for a rehabilitation application or in a sport motion learning application. This application is based on my research on human motion analysis. It allows a “teacher” to record a new movement using a Kinect v2. Then a “student” can try to perform the same gesture, and a corresponding score is given. In addition, the user can visualize a feedback where its body parts are colorized according to a color map to highlight errors.