Fig. 3
From: Detecting the symptoms of Parkinson’s disease with non-standard video
Our data processing procedure has four main steps: A manual segmentation of full-length clinical assessment videos into individual tasks; B participant keypoints detection using PosePipe and PIXIE to generate timeseries data; C calculation of joint kinematics and other timeseries signals from the detected keypoint timeseries; and D extraction of advanced features of functional motor performance from the joint kinematics data