AI Golf Coach: A Pose-Based Golf Coaching System using Artificial Intelligence and Computer Vision

Abstract

In the sport of golf, several main mechanics in each successful golf swing bring consistency [1]. In addition, there are specific nuances in the golf swing that may be difficult for recreational players to analyze and detect. The main question is how an app can potentially detect these main techniques and compare them when inputting two different golf swings. Guided by Mediapipe and cv2 technologies, we have built software to analyze two golf swings simultaneously, and to compare them with one another, all into a smartphone application [2][3]. This paper will discuss key technologies that were utilized to create the program, the various challenges that were faced in the process of creating the software, along with our methodologies [4]. The end result was an application that would detect various angles created based on certain golf positions, and then utilize them to compare with another video, whether that be a professional player or a friend’s golf swing. Using a Python backend and a Flutter-based frontend, the program can be utilized through a mobile device, and thus makes the program accessible to the public [5]. With the publishing of this application, users can start to deconstruct their own golf mechanics without the complications and the high fee that is generally charged for one-on-one coaching.