2018
The following work was inspired and guided by Drematrix.
Thanks to the improving healthcare system, the life expectancy of Parkinson’s patients is also increasing. Regular exercise can help these patients maintain their motor skills for as long as possible and aims to minimize the need for medication, surgery and walking aids.
Drematrix’s project is aimed at improving the daily living conditions of Parkinson’s patients: One for the patient as a training option in the home environment, the other as a monitoring tool for therapists in rehabilitation clinics. It is intended to be adaptable, so it must be based on the personal needs of Parkinson’s patients and aims to combine an appealing presentation with an accurate assessment of motor skills.
For the project the data were recorded with a Xbox Kinect, using the integrated IR depth sensor.
For my work/tests I used the marker less motion capture facility at my university to capture motion data. Therefore it was important to create the same condition for every recording. The cameras had to be calibrated before shooting.
Inside Unity the raw data was cleaned and processed with a OneEuroFilter. OpenPose was then fed with the data. OpenPose is a realtime post estimation tool (https://github.com/CMU-Perceptual-Computing-Lab/openpose)
Open Pose: To minimize jitter and lag when tracking human motion, the two parameters (fcmin and beta) can be set using a simple two-step procedure. More information can be found on the following website: https://cristal.univ-lille.fr/~casiez/1euro/ .
Technologies: Unity, MakeHuman