The goal of 3D Media In Sports experiment at CAR is to investigate the usefulness of 3D information in high performance sports training centres, which is proved in weightlifting and cycling. 3D information is a key element for accurate and productive training in many aspects. For instance, information that in the past was captured either solely by the trainer and interpreted thanks to his/her experience (such as knee angles, torso angles, etc.) or from annoying wearable devices that can affect the athlete’s performance, can now be automatically gathered by non-intervening sensing devices and processed in real-time so as to provide useful feedback to the user (trainer and/or athlete). By being able to accurately manipulate this information a whole new world of opportunities emerges in the Future Media Internet (FMI). For instance, since 3D biomechanical information can be streamed over the Internet seamlessly and accurately, sports community dedicated social media can benefit of the use of such information to provide new 3D media services to their users.
The Microsoft Kinect was a breakthrough device for the easy capturing of 3D information. This fact led to the enormous penetration of the FMI ideas to a very wide audience, yet many inefficiencies remain unsolved. With the aforementioned experiment, EXPERIMEDIA developed sophisticated algorithms to cope with these inefficiencies, especially for the demanding high performance sport cases. The algorithms provide information correction from inaccurate depth estimation, constrain the Kinect’s calculated human skeleton data to physical poses to enhance reliability, as well as many more enhancement that are suitable for sports biomechanical measurements. Moreover, novel algorithms were developed to calculate several biomechanical measurements for two use cases (i.e., weightlifting and cycling). In the case of weightlifting the knee angle of the weightlifter, the velocity of the weightlifting bar and the angle of the weightlifting bar with the horizontal axis (the one parallel to the floor) are calculated. In the case of the cycling, the cyclist frontal area, the arm-torso angle, the horizontal axis-torso angle and the forearm-arm angle are calculated. Therefore, EXPERIMEDIA clearly offers a unique combination of 3D based biomechanical information with FMI, and thus, paving the way for more novel application domains.
Our algorithms provide the means to move high performance sports training a step beyond. New technologies for 3D data acquisition, 3D information enhancement/correction and biomechanical measurements have been developed and applied in two use cases (i.e., weightlifting and cycling). The end-users (e.g., athletes and/or trainers) have much appreciated the non-intervening/invasive nature of the proposed solution. Moreover, they considered that the results from the algorithms of the biomechanical measurements, both in cycling and weightlifting use cases, were fair in terms of accuracy and also informative and useful to their training sessions.. High performance sports training centres such as CAR benefit of the use of such technologies and can provide better services to their athletes and trainers in terms of more productive training sessions.