We developed a solution which combines views from a multi camera setup resulting in 3D images.
From this 3D visual hull, relevant information such as presence, position, speed, acceleration and the kinematic model can be derived.
This information is then coupled to data analysis tools leading to enriched decision making data for coaching, scouting and refereeing.
Examples of our latest work at the Sport Science Laboratory – Jacques Rogge can be found here:
Hardware: off-the-shelf mid-range IP cameras, calibration tools, a custom designed data processing and storage solution and visualization equipment. The number of cameras depends on the complexity of the scene you want to observe (e.g. amount of people to track, their proximity to one another and the amount of scene occlusion).
Software: state-of-the-art FG/BG segmentation algorithms, the visual hull algorithm including occlusion management and analytics tools.
Statistics on the tracking of players in sports games prove to be very valuable. However, current solutions are very expensive and are not always applicable at game time.
Marker based systems can only be used reliably in indoor conditions. Radio based transmitters have a varying reliability depending on the surroundings. Both need to be applied to the person’s body, which is often undesired or impossible.
In many sports, wearables at game time are not allowed because of regulations, safety issues or performance-interferencing reasons.
Markerless motion capturing makes wearables redundant while performance analysis is still possible, even at game time.
A remarkable aspect of the technology is the intelligent occlusion management. This effectively overcomes interference to the person’s 3D approximation originating from objects blocking the view of the observed person. The latter is often overlooked in current technologies, significantly limiting their use.
The team is very responsive for innovation and business development feedback. As strong believers in lean innovation, they are convinced that assumptions related to the technology, the market it can possibly be applied in, etc. should be continuously tested with end-users.
A possible application of the technology, that the team is working on for the moment, is applying this markerless motion capturing technology in basketball shooting analytics.