Badminton data analytics in joint angles and distances between joints: assessing the ai advantage over human expertise
Background: Video analysis in sports has grown, with markerless motion capture systems like BlazePose offering better accuracy and convenience than marker-based systems. Joint angles and distances are crucial for assessing skills and injury risk in sports like badminton. BlazePose's technology...
Saved in:
Main Author: | |
---|---|
Other Authors: | |
Format: | Final Year Project |
Language: | English |
Published: |
Nanyang Technological University
2023
|
Subjects: | |
Online Access: | https://hdl.handle.net/10356/171836 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Nanyang Technological University |
Language: | English |
Summary: | Background: Video analysis in sports has grown, with markerless motion capture systems like BlazePose offering better accuracy and convenience than marker-based systems. Joint angles and distances are crucial for assessing skills and injury risk in sports like badminton. BlazePose's technology has revolutionized motion capture, allowing precise pose tracking for real-time performance analysis. Aim: The aim of this study was to validate the accuracy of Mediapipe BlazePose Ghum 3D model in measuring joint angles and distances between joints. Method: Three male participants were recruited to perform 41 different badminton shots. Qualisys (marker-based system) as well as Mediapipe were used to output joint coordinates in 3D. Joint distances and angles were then calculated using trigonometry. Results: Results found that most joint angles had high nPVI and RMSE values, especially for angles in a different plane from the camera (Mediapipe). Distances between joints were inconsistent with some having great accuracy with nPVI less than 10% but some close to 90%. Markerless motion capture system showed substantial accuracy but limited sensitivity when assessing complex joint angles and mid-range angles and distances. Conclusion: This research sheds light on the model's applicability in distinguishing nuanced badminton shots, where high sensitivity is crucial. Hence, users should judiciously consider the technology's suitability for their specific requirements in the realm of badminton or similar domains, keeping in mind its strengths and limitations. |
---|