Manual Marker-Less Digitizing:
Dr. Gideon Ariel and Alan Blitzblau (who later co-founded Bio Kinetics) collaborated in the 1970’s to develop the first three-dimensional sports motion analysis system. This system employed manual digitizing from multiple simultaneous film images (and eventually video images) to accurately measure and model the body mechanics of athletes in actual competition in a wide range of sports. Manual digitizing from multiple calibrated cameras has been the accepted process for measuring human movement world-wide for almost five decades. Manual digitizing has been used in the Olympic Games since 1972 and thousands of articles have been published using this technique. In addition, recent advancements in software, camera technology and monitor resolution have improved the accuracy of manual digitizing. This is the process Bio Kinetics uses to perform all motion analysis at this time.
Shows manual digitizing from the front and side view.
Automated Marker Digitizing:
Automated laboratory marker tracking has been available since the mid 1980’s. This process is reasonably accurate and has a tremendous advantage in speed. However, under certain conditions, it is not as accurate as manual digitizing. In automated digitizing the computer measure’s locations of the markers not the body’s joint centers. Consider a marker placed on the side of a pitcher’s shoulder that is in line with the joint center. Now, when he lifts his arm to throw, the marker is no longer aligned with the joint center. Markers are attached to the subject’s skin, not to the joint centers. During body movement skin migrates relative to the underlying bones and joints, and thus the marker would migrate along with the skin. Furthermore, the anatomical joint center can also migrate within the body because a joint is not a perfect hinge – the shoulder is a good example of this. There are other problems with markers as well. While attending a convention, one major 3D marker system manufacturer told us that it was important to capture the first throw, before the athlete warms up and starts to sweat, so that the markers don’t fall off. That might not be the best idea for a pitcher! Skin markers always move relative to the underlying bone structure. This movement is minimal during slow activities, and thus marker-based systems work well for these. However, activities that involve very high acceleration of body segments, such as the hitting and pitching, can result in a significant movement of skin markers relative to the underlying bones (picture Clint Eastwood and Tommy Lee Jones on the G-Sled in the movie Space Cowboys where their faces are all deformed during high-G acceleration). In situations such as this a marker-based system is much less accurate than a well-digitized manual system. During manual digitizing a skilled digitizer can use body landmarks to visualize the location and orientation of the joint center more accurately.
Marker based systems also have a significant disadvantage in that they can NOT be used during live competition. What a player does biomechanically in a controlled environment or laboratory setting is not the same, in fact does not accurately resemble what players do biomechanically during an actual game situation. Bio Kinetics does NOT use marker-based digitizing because it cannot be used during live game competition.
Shows a pitcher with markers attached to various positions on the body for automated three-dimensional motion capture.
Marker-Less Tracking:
Bio Kinetics has been following the progress of marker-less technology over the last decade. This tracking method uses artificial intelligence to analyze body images in video frames in order to estimate the location of body joints. (Estimating anything with a player’s career at stake may not be the best idea) University studies have tested live subjects against several different backgrounds, lighting, and with various clothing with some success. In the future this may prove to be a promising technique, but for now this technology has only been demonstrated to be reliable in a laboratory setting with slower and less rotational movements. In a baseball stadium, during live competition,
with various backgrounds and lighting conditions, there are NO existing marker-less capture technologies capable of capturing game-speed motion with the necessary degree of accuracy for analysis. Bio Kinetics will NOT use this technology until its accuracy in outdoor “large area” settings such as a stadium is established in the academic / scientific community.
Sensors:
Wearable and implement-placed wireless motion sensors have now made their way into baseball. As with marker-less tracking, university studies have shown these devices can accurately capture many types of movement in a laboratory setting but they require time-consuming setup and special instrumentation to use, and they have a limited range of transmission. They have little value in evaluating true athletic performance at this time as they cannot be used in live competition.
Pitch Out
Since marker-less tracking and sensor tracking companies have made their way into baseball, several organizations have sent us sample data to be compared to our established and validated process of measuring human movement (manual digitizing). Thus-far, marker-less tracking has not done a good job in capturing complex, rapid motion as is seen in the pitching delivery. A professor who was reviewing some of the data comparisons, jokingly commented “if that was his DNA he would be a dolphin”. We have noticed some degree of improvement over time, however that presents another problem! As they continue adjust their AI models that transform a silhouette into an estimate of body joint locations (instead of actual digitized measurements of body joints) the data obtained from previous analyses can no longer be used in comparison as their measuring device, and thus the measured data itself has changed.