Kinovea & Open Pose for Display of Tennis Strokes

Chas Tennis


I just read a description of a new version of Kinovea, Kinovea 9.1. Kinovea is free and open source.
(I don't know what .NET 4.8 is. ?)

Kinovea has added a new capability using a machine learning application, Open Pose. Kinovea can import Open Pose files and use them.

These points of the body, often near or at joints, are tracked in 2D single camera views.

Information from a similar machine learning application, OpenCV

Normally, 3D 'motion capture systems' use very expensive multi-cameras and computers and require special expertise. This Open Pose and similar machine learning technologies use a typical video from any single camera and locate body joints and orientations without markers. It can show some very informative biomechanical angles and their timing for tennis strokes as in the Wimbledon backhands above.

For comparing many angles in side-by-side videos, this technology looks very interesting and fast. It can even analyze more than one athlete performing in a frame. It may be limited to locating the main joints and have some limitations, for example, especially on seeing joint rotations where the bone rotates but does not change location.
Last edited: