Golf lifestyle platform Kim Caddy announced on the 14th that a paper on golf swing analysis technology developed by its artificial intelligence (AI) research team will be presented at the Sports Vision Workshop (CVSports) of 'CVPR 2025 (IEEE/CVF Conference on Computer Vision and Pattern Recognition),' a world-class computer vision conference.
The paper is titled 'CaddieSet: A Golf Swing Dataset with Human Joint Features and Ball Information' and introduces a dataset for analyzing golf swings based on actual users. Researchers Jae-rim Choi and Seo-young Hong from the Kim Caddie AI Team, as well as researchers from the computer science department of Dongguk University's Professor Woo-jin Lee's lab, and Professor Ho-gi Kim from Chung-Ang University jointly participated in the research.
According to the paper, 'CaddieSet' was built based on video data of actual golfers collected through the Kim Caddie platform. It divided the golf swing motion into eight stages, and calculated a total of 15 motion indices based on the joint position information extracted from each stage. It is characterized by being configured to quantitatively analyze the relationship between motion and result by linking physical data such as the direction, rotation axis, and speed of the ball that occurred in the same swing.
This study is evaluated as the first quantitative analysis case using general user-based golf movement data. The dataset used for AI model learning improved the accuracy of ball trajectory prediction compared to existing image-based analysis, and has the function of AI explaining the cause of swing errors. For example, AI is designed to infer the cause of the ball bending in a certain direction from the address posture or lack of weight transfer, and suggest a direction for correction.
Kim Caddy will demonstrate the technology to the general public at the 'AWS Summit Seoul 2025' event scheduled to be held in May. Through the experiential booth called 'Nova Golfriend', users will be able to experience AI-based swing diagnosis, golf MBTI test, tee-sticking game, etc., and various souvenirs will also be provided.
Professor Lee Woo-jin of Dongguk University said, “This paper is significant in that it goes beyond technical achievements and demonstrates the potential of an AI-based analysis system that can be applied to actual industrial sites,” adding, “We plan to conduct follow-up research and commercialization efforts in parallel so that AI technology can provide practical value to the golf industry as a whole.”
- See more related articles
You must be logged in to post a comment.