Background: Video analysis can be not only used to obtain the kinematical data of weightlifting movements for scientists, but also served as a way of feedback for coachers and
weightlifters. However, post- processing is needed to obtain the kinematical data, so it is impossible to provide coachers the video feedback with relevant kinematical data in real-time. Therefore,
the purpose of this study was to develop a real-time feedback system (RTFS), which can provide the coachers and weightlifters kinematical data and video information.
Methods: A Kinect sensor was used as the hardware to capture the depth data and RGB video, the methods of the pattern recognition and algorithm were established, and corresponding
software was developed using Kinect SDK v1.5 and C++ programming language to identify the barbell and calculate the data of barbell center of mass (COM) in three dimensions. In order to check the
reliability of the data from RTFS, an experiment was carried out to compare the data from RTFS and that from 3D analysis based on video(SIMI Motion).
Results: The statistical analysis (T-Test) showed that there were no significant differences existing between the displacements (x,y,z) of barbell COM from SIMI Motion system and
RTFS, this means that the data exported by RTFS in real-time are reliable and credible, and can be used to calculate the other kinematical parameters. The real-time feedback system developed in
this study was mainly designed for coachers to diagnose and evaluate the technique of weightlifters in daily training, so this system has following specific features:
1 Without any marker, set up and use the system very easy
2 The system starts and stops capturing automatically by means of judging the bar left from the ground and dropped to the ground
3 After the bar dropped to the ground, the video captured, data and curve, such as the maximal height of the bar, the trajectory of the bar were replayed on LED screen immediately
4 Pre-defined evaluation protocols for snatch and clean & jerk can be automatically chosen by judging the characteristic of bar movement
5 Video and kinematical data can be saved directly to database for the further comparison and analysis
6 The system has two screens, one is touch screen for easy operation, the other one is large screen for showing the feedback of results
Conclusions: The system developed in this study is an integrated system, which has the comprehensive systematic functions including the depth images & video recording, intelligent
pattern recognition, data processing and saving, results feedback and exporting. It was designed particularly for coaches in weightlifting training to improve the performance executed by lifters. The
innovation of this system is to realize the feedback in real-time, which makes the feedback more and more effective than before. China National Weightlifting Team has been using this system in their
daily training, and some improvements will be made to meet the demands for practical applications.