Authors: Dimitrije Cabarkapa1, Andrew C. Fry1 and Eric M. Mosier2
- Jayhawk Athletic Performance Laboratory, University of Kansas, Lawrence, KS
- Northwest Missouri State University, Maryville, MO
Dimitrije Cabarkapa, MS, CSCS, NSCA-CPT, USAW
1301 Sunnyside Avenue, Lawrence, KS 66045
University of Kansas
Phone: +1 (785) 551-3882
Validity of 3-D Markerless Motion Capture System for Assessing Basketball Dunk Kinetics – A Case Study
Basketball is one of the most popular international sports, but the current sport science literature does not directly address on-court performance such as force and power during a game. This case study examined the accuracy of a three-dimensional markerless motion capture system (3-D MCS) for determining the biomechanical characteristics of the basketball dunk. A former collegiate (NCAA Division-I) basketball player (age=26 yrs, height=2.08 m, weight=111.4 kg) performed 30 maximum effort dunks utilizing a two-hands, no-step, two-leg jumping approach. A uni-axial force plate (FP) positioned under a regulation basket sampled data at 1000 Hz. Additionally, a 3-D MCS composed of eight cameras placed 3.7 m high surrounding the recording area collected data at 50 Hz, from which ground reaction forces were derived using inverse dynamics. The dunks were analyzed by both systems for peak force and peak power. Peak force (X±SD) was similar (p<0.05) for both systems (FP= 2963.9±92.1 N, 3-D MCS= 3353.2±255.9 N), as was peak power (FP= 5943±323, 3-D MCS= 5931±700 W). Bland-Altman plots with 95% confidence intervals for both force and power indicated all measurements made with the 3-D MCS accurately assessed peak force and peak power during a basketball dunk as performed in the current study. These data provide strength and conditioning professionals with a better understanding of the magnitude of forces and powers that athletes experience during a basketball game, as well as validate use of a novel technology to monitor athletes’ progress and optimize overall athletic performance.
Key words: jump, basketball, force, sport
Basketball is one of the most popular international sports, but the current sport science literature does not directly address on-court performance such as force and power during a game. Force plate technology is highly used and many practitioners currently consider it the gold standard for assessing peak force and power outputs. However, it has certain limitations including lack of portability and time-consuming data analysis. Many motion capture systems allow the collection of kinetic and kinematic variables of human motion using marker-based methods. The set-up and data collection procedure using this technology consumes a significant amount of time and requires accurate and consistent placement of markers. Recently, markerless motion capture systems have been developed to address this issue. Mündermann et al. (4) have emphasized the importance and necessity for substituting old marker-based motion capture systems with markerless motion capture systems that are highly portable, simple, time efficient and are able to observe human motion in a three-dimensional perspective.
A previous report on the accuracy and repeatability of markerless motion capture technology indicated that lower body kinematic values were not different when compared to a marker-based motion capture system (7). In addition, Perrot et al. (5) revealed that a markerless motion capture system reported similar joint angles as a marker-based motion capture system when knee flexion and single leg squat exercises were observed. However, it is important to note that the field of sport science and engineering is still in development. While it is common to find studies indicating high consistencies between a marker-based and markerless motion capture systems, some of the systems examined did not demonstrate complete agreement. When lower body joint rotations were examined utilizing marker-based versus markerless motion capture systems, researchers found that flexion/extension movements and hip abduction/adduction movements produced consistent results compared to a marker-based motion capture system (6). On the other hand, measurements related to accuracy of internal/external rotations of the knee and inversion/eversion of the ankle were less reliable (6).
Recently, Cabarkapa & Fry (1) used force plate technology to examine ground reaction forces for some of the most commonly performed basketball dunking approaches. However, this type of analysis must be performed in the laboratory or on specifically constructed basketball court. The use of recently developed markerless motion capture systems capable of quantifying sport performance would make these assessments much more practical, but these measurements must be valid and reliable (8). It was recently reported that a markerless motion capture system accurately quantified ground reaction forces when compared to a force plate equipment (2). Hence, the purpose of this case study was to examine the accuracy of a three-dimensional markerless motion capture system (3-D MCS) when compared to a force plate, serving as the gold standard, for determining the biomechanical characteristics of one of the most powerful activities in the sport of basketball, the dunk.
A former National Collegiate Athletic Association (NCAA Division-I) collegiate basketball player (age = 26 yrs, height = 2.08 m, weight = 111.4 kg, playing experience = 12 yrs) performed 30 dunks using a two-hands, no-step, two-leg jumping approach. Upon reading and signing the informed consent form, the subject completed a health history questionnaire as well as a video authorization form. All procedures performed in this research study were approved by University’s Institutional Review Board.
A uni-axial force plate (0.91m X 2.44m; Rice Lake Weighing Systems, Rice Lake, WI, USA) was positioned under a regulation basket, with force sampled at 1000 Hz with a BioPac data acquisition system (Goleta, CA, USA). Additionally, a 3-D MCS (DARI, Lenexa, KS, USA) composed of eight identical cameras placed 12 feet (3.7m) high surrounding the recording area collected data at 50 Hz. The basket was placed at the designated height of 10 feet (3.05 m) corresponding to well established basketball rules. For this study a basketball size 7 (29.5 inches diameter and 22 oz) was used. The subject wore dri-fit clothes in order to enhance accuracy of the 3-D markerless motion capture device. Additionally, the recording area was covered with green colored floor pads to avoid light reflection that could interfere with data collection. The complete equipment set-up is presented in Figure 1.
Figure 1 – Experiment set-up. Small arrows – motion capture system cameras, large arrow – force plate.
After completing a dynamic warm-up, the subject performed 30 dunks with maximum effort each time using a no-step approach, a two-legged take-off, and holding the basketball in both hands. For each dunk, both the force plate and the 3-D MCS recorded data simultaneously. In order to minimize fatigue, the subject actively rested for two minutes between each trial. Before each dunk jump, the subject was instructed to stand still on the force plate for several seconds as an essential step for determining the initiation of the countermovement portion of the dunk jump, very similar to a vertical jump.
Resulting ground reaction forces from the force plate were analyzed for both peak force (N) and peak power (W). The 3-D MCS software inversely derived ground reaction forces for each dunk, which were also analyzed for concentric peak force and concentric peak power. The force plate data, recorded at 1000 Hz, was converted to 50 Hz to correspond to the 3-D MCS sampling frequency. To synchronize both force curves, we aligned the peak forces obtained from both the force plate and 3-D MCS system measurements as presented in the Figure 2. The starting point, indicating the initiation of the dunk motion, was set as the initial change in ground reaction force value compared to the subjects’ body weight determined during the subjects’ motionless standing on the force plate. The end point was set as three data points (duration of 0.06 s) where both force plate and 3-D MCS ground reaction force curves displayed zero values, indicating that the subject was airborne. Peak power was derived as the product of force and velocity for both the force plate and the 3-D MCS.
Figure 2 – An example of synchronized concentric ground reaction force curves for both the force plate (FP) and the three-dimensional markerless motion capture system (3-D MCS).
Peak force and peak power for both testing methods were compared using dependent t-tests (p < 0.05). Bland-Altman plots using 95% confidence intervals (CI) were used to depict agreement of the 3-D MCS when compared with the gold standard force plate data. Finally, in order to compare the resulting force-time curves from both the force plate and the 3-D MCS, all data points for each individual dunk were compared using intra-class correlation coefficients (ICC).
Peak forces (X±SD) for the force plate (FP) and the 3-D MCS were FP = 2963.9 ± 92.1 N and 3-D MCS = 3353.2 ± 255.9 N, and were not significantly different. Also, peak powers were FP = 5943 ± 323 W and 3-D MCS = 5931 ± 700 W, which were also not significantly different. Bland-Altman plots for both force and power indicated that all measurements made with the 3-D MCS were well within the 95% confidence interval when compared to the FP derived data, which are presented in Figures 3 and 4. ICC values for 29 out of 30 dunks ranged between 0.920-0.992, with only one trial exhibiting a lower value (ICC=0.793).
Figure 3 – Bland-Altman plot for peak force, solid line = line of agreement, dashed lines = 95% CI, dotted line = regression line. FP = force plate, 3-D MCS = three-dimensional markerless motion capture system.
Figure 4 – Bland-Altman plot for peak power, solid line = line of agreement, dashed lines = 95% CI, dotted line = regression line. FP = force plate, 3-D MCS = three-dimensional markerless motion capture system.
Innovative 3-D MCS technology such as used in the present study, when compared to the gold standard force plate-derived data, was shown to be a valid method for assessing the kinetics of a basketball dunk. Both methods demonstrated acceptable agreement within 95% confidence intervals for peak force and power, as well as consistency of measurements during multiple dunk trials. Peak ground reaction forces obtained in this study agree with the findings of Cabarkapa & Fry (1) findings for kinetic characteristics of a two-leg stationary dunking approach. There are currently several manufacturers of markerless motion capture systems commercially available, but their validity compared to marker-based systems for various sporting activities have not been established. One related project recorded American football players during their preseason training. The markerless motion capture system used accurately assessed the majority of body movements kinematics when compared to a marker-based system, although rotational movements were less accurate (3). In another example, Schmitz et al. (7) observed the squat as a fundamental body movement, and found that markerless motion capture system technology accurately assessed kinematic variables when compared to a marker-based system. In the present study, the kinetic variables of peak force and peak power for a basketball dunk were validly and easily collected by an advanced 3-D MCS technology. Further research may now be possible to validly and reliably determine the kinematic and kinetic components of other basketball skills, or other sports specific activities.
Based on the results obtained in the present case study, we conclude that the innovative 3-D MCS used in the study accurately assessed peak force and peak power during a basketball dunk. While the force plate served as the gold standard, the 3-D MCS allows greater freedom of movement, and the accompanying software produced rapid data analysis for peak force and peak power during the basketball dunk.
APPLICATIONS IN SPORT
The measured kinetic properties for the basketball dunk can also provide coaches and athletes with valuable information concerning the magnitude of forces and powers experienced during a basketball game. It is likely that technology such as used in this case study can help coaches and athletes understand and evaluate a variety of sports activities. Through decreased testing times and immediate feedback, coaches and athletes can make rapid modifications of sport techniques and training methods to optimize athletic performance, which can aid in developing optimal strength and conditioning program designs.
- Cabarkapa D., & Fry, A. C. (2019). Ground reaction forces during a basketball dunk. Slovak Journal of Sport Science, 4(1-2).
- Fry, A. C., Herda, T. J., Sterczala, A. J., Cooper, M. A., & Andre, M. J. (2016). Validation of a motion capture system for deriving accurate ground reaction forces without a force plate. Big Data Analytics, 1(1), 11.
- Kotsifaki, A., Whiteley, R., & Hansen, C. (2018). Dual Kinect v2 system can capture lower limb kinematics reasonably well in a clinical setting: concurrent validity of a dual camera markerless motion capture system in professional football players. BMJ Open Sport & Exercise Medicine, 4(1), e000441.
- Mündermann, L., Corazza, S., & Andriacchi, T. P. (2006). The evolution of methods for the capture of human movement leading to markerless motion capture for biomechanical applications. Journal of Neuroengineering and Rehabilitation, 3(1), 6.
- Perrott, M. A., Pizzari, T., Cook, J., & McClelland, J. A. (2017). Comparison of lower limb and trunk kinematics between markerless and marker-based motion capture systems. Gait & Posture, 52, 57–61.
- Sandau, M., Koblauch, H., Moeslund, T. B., Aanæs, H., Alkjær, T., & Simonsen, E. B. (2014). Markerless motion capture can provide reliable 3D gait kinematics in the sagittal and frontal plane. Medical Engineering & Physics, 36(9), 1168–1175.
- Schmitz, A., Ye, M., Shapiro, R., Yang, R., & Noehren, B. (2014). Accuracy and repeatability of joint angles measured using a single camera markerless motion capture system. Journal of Biomechanics, 47(2), 587–591.
- Vincent, W. J., & Weir, J. P. (1994). Statistics in Kinesiology 4th Edition. Human Kinetics.