Skip to main content
. 2019 Jun 7;19(11):2604. doi: 10.3390/s19112604

Figure 1.

Figure 1

Motion skeleton estimates from two types of motion extraction SDKs. Upper row: The motion skeleton captured by Kinect v2.0. (AD) Four frames in the a motion clip and (E) all of them in a single image. Bottom row: The motion skeleton captured by Intel RealSense D435 with the Nuitrack SDK. (FI) Four frames in a motion clip and (J) all of them in a single image. This demonstrates that the respective skeleton estimates vary significantly regarding the bone length during the motion.