Pen Spinning Hand Movement Analysis Using MediaPipe Hands
This work addresses the need for objective performance evaluation in pen spinning competitions, though it is incremental as it builds on existing computer vision tools.
The authors tackled the problem of subjective evaluation in pen spinning competitions by developing a system using MediaPipe Hands and OpenCV to analyze hand movements, achieving accurate results through adjustments in saturation and brightness to improve recognition accuracy.
We challenged to get data about hand movement in pen spinning using MediaPipe Hands and OpenCV. The purpose is to create a system that can be used to objectively evaluate the performance of pen spinning competitions. Evaluation of execution, smoothness, and control in competitions are quite difficult and often with subjectivity. Therefore, we aimed to fully automate the process by using objective numerical values for evaluation. Uncertainty still exists in MediaPipe's skeletal recognition, and it tends to be more difficult to recognize in brightly colored backgrounds. However, we could improve the recognition accuracy by changing the saturation and brightness in the program. Furthermore, automatic detection and adjustment of brightness is now possible. As the next step to systematize the evaluation of pen spinning using objective numerical values, we adopted "hand movements". We were able to visualize the ups and downs of the hand movements by calculating the standard deviation and L2 norm of the hand's coordinates in each frame. The results of hand movements are quite accurate, and we feel that it is a big step toward our goal. In the future, we would like to make great efforts to fully automate the grading of pen spinning.