Evaluation of Pose Tracking Accuracy in the First and Second Generations of Microsoft Kinect

Abstract

Microsoft Kinect camera and its skeletal tracking capabilities have been embraced by many researchers and commercial developers in various applications of real-time human movement analysis. In this paper, we evaluate the accuracy of the human kinematic motion data in the first and second generation of the Kinect system, and compare the results with an optical motion capture system. We collected motion data in 12 exercises for 10 different subjects and from three different viewpoints. We report on the accuracy of the joint localization and bone length estimation of Kinect skeletons in comparison to the motion capture. We also analyze the distribution of the joint localization offsets by fitting a mixture of Gaussian and uniform distribution models to determine the outliers in the Kinect motion data. Our analysis shows that overall Kinect 2 has more robust and more accurate tracking of human pose as compared to Kinect 1.

Classification of Gait Anomaly due to Lesion Using Full-Body Gait Motions

“…For example, it is reported that RMSEs for knee and hip joints in a gait sequence are 28.5 degrees and 11.8 degrees, respectively, in [18]. More comprehensive error analysis was conducted in [16]. By reducing such error by employing a human motion prior for pose tracking [31]- [33], it is expected that the identification performance of the proposed method is improved.…”

“…The depth sensor is able to robustly reconstruct the 3D pose (i.e., 3D positions of joints) of a person of interest [14], [15] rather than a conventional RGB camera. A Kinect V2 sensor was used in our experiments, and its accuracy is better than that of Kinect V1 [16], [17] for several tasks related to gait analysis [18], [19].…”

An Automated Method to Extract Three-Dimensional Position Data Using an Infrared Time-of-Flight Camera

“…For long-range camera distances (i.e., 2 m < camera distance < 4 m), systematic thigh angle differences increased with camera distance. This reflects known depthaccuracy limits of depth-cameras [8] as well as decreasing resolution of tracking targets (i.e., markers). However, close-range measurement errors were not related to camera distance; rather, an optimal camera-distance emerged (Table 1).…”

“…Infrared images captured (~30 Hz) by an Xbox One Kinect (Microsoft, Redmond, USA) are first converted to logical images (based on a threshold), allowing retro-reflective markers to be tracked (e.g., Figure 1a). Depth data-based on the phase-shift distance of modulated infrared light [8]-cannot be derived from marker locations directly as retro-reflective material scatter infrared light. Therefore, image locations of retro-reflective markers (e.g., Figure 1a) are used to track three coplanar image targets (e.g., red, green and blue tracking crosshairs highlighted by 'L' shape in Figure 1a) within infrared images.…”

“…The advent of infrared time-of-flight depth-cameras (e.g., Xbox One Kinect), which provide higher resolution sensors with increased depth measurement fidelity [8], might yield improved accuracy, addressing previous limitations [7]. Therefore, the use of infrared time-of-flight depth cameras might expand potential applications of depth-cameras for healthcare specialists.…”