Golf Swing Motion Analysis Using Multi-Scene Posture Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing golf swing analysis systems fail to perform accurate evaluation across multiple scenes, such as address and top of swing, by considering the posture of a player's body parts and club position.
Innovation Solution
A motion analysis system and method that extracts and identifies positions of body parts and golf club in different swing scenes, using skeletal and object recognition technologies, to evaluate posture accurately by comparing positions across scenes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If evaluation is performed using only a single scene (e.g., impact), then the analysis system is simple to operate, but the evaluation accuracy is insufficient
Solution Approach 1:
The golf swing is divided into multiple distinct scenes (address, takeaway, top of swing, impact, follow-through), with each scene analyzed separately using dedicated frame images and evaluation criteria. This segmentation allows comprehensive multi-scene evaluation while maintaining manageable system complexity through modular scene-specific analysis.
Solution Approach 2:
The system transitions from single-scene analysis to multi-scene analysis by adding the temporal dimension of swing phases. By analyzing multiple scenes sequentially and comparing key point positions across scenes, the system achieves comprehensive evaluation without overwhelming complexity through structured phase-by-phase processing.
2Measurement precision
If multiple scenes are analyzed to improve evaluation accuracy, then more comprehensive feedback is provided, but the processing time increases
Solution Approach 1:
The system performs preliminary extraction of frame images representing each swing scene and pre-identification of key point positions (body parts and club positions) in advance. This preliminary processing of multiple scenes allows the evaluation to be conducted efficiently by comparing pre-identified positions rather than processing raw video data in real-time during evaluation.
Solution Approach 2:
The system creates simplified representations (copies) of the golf swing by extracting key frame images and identifying critical key points in each scene. Instead of analyzing the entire continuous video, the system works with these copied essential elements, significantly reducing processing time while maintaining evaluation accuracy through focused analysis of critical moments.
3Measurement precision
If detailed position identification of body parts and club is performed, then evaluation precision is improved, but the difficulty of detecting and measuring increases
Solution Approach 1:
The detection task is segmented by scene (address, takeaway, top of swing, impact, follow-through), with each scene's frame image processed independently for key point identification. This segmentation reduces detection complexity by focusing on specific moments rather than continuous analysis, while achieving high position identification accuracy through scene-specific optimization.
Solution Approach 2:
The system applies different detection and identification approaches tailored to each swing scene's characteristics. By optimizing key point identification methods for specific local conditions of each scene (e.g., club position at address vs. impact), the system achieves high measurement precision while managing detection complexity through localized specialized processing.
Data Source
AI summary
A frame image extraction module extracts a first frame image representing a first scene and a second frame image representing a second scene. An object position identification module identifies a position at which a first object is shown in the first frame image and a position at which a second object is shown in the second frame image. An evaluation processing execution module evaluates a posture of a person in the second scene based on the position at which the first object is shown in the first frame image and the position at which the second object is shown in the second frame image.