Golf players often rely on experience and external tools to guide their gameplay. The goal of this project was to develop an augmented reality application that provides real-time guidance to golfers through wearable AR glasses, helping both professionals and amateurs improve their performance on the course.
The application required advanced computer vision capabilities to understand the environment and provide real-time insights during gameplay. The system needed to track golf balls and player stance while running efficiently on lightweight AR hardware.
- Developing computer vision models capable of recognizing golf balls and player stance in real time
- Processing spatial data to understand the golf course environment in three dimensions
- Running machine learning models on lightweight AR hardware with limited processing capacity
- Delivering a smooth augmented reality experience without distracting players during gameplay
- Supporting administrators in managing providers, users, and educational content
Technology Stack
-
Architecture
Unity -
Hardware
TCL RayNeo XR Glasses, Android companion device -
SDK
OpenXR -
Computer Vision
YOLOv8 and Google MediaPipe -
Cloud Infrastructure
AWS ECS with auto-scaling -
Integration
Zoom API, Google Calendar APIs
Virstack developed an augmented reality system that combines computer vision, machine learning, and spatial awareness to provide real-time golf guidance through wearable AR glasses.
Computer Vision Object Tracking
Machine learning models trained on thousands of golf images enable real-time recognition and tracking of golf balls and player positioning.
Augmented Reality Gameplay Guidance
The system uses course data and player history to provide recommendations and guidance during play.
Spatial Awareness and 3D Rendering
A spatially aware 6DoF system allows the application to understand the environment around the player and render AR guidance in real time.
Mobile Companion Integration
A companion mobile device connects to the AR glasses to support the user interface and data processing.
Voice Guidance and Minimal Interface
Voice prompts and a simplified interface allow players to focus on their game without distractions.
The platform created a centralized ecosystem for parenting support and lactation consultation.
-
A working MVP enabling demonstration of the AR golf guidance concept
-
A lightweight computer vision model capable of running on AR hardware
-
Cost-efficient development of a prototype without large training infrastructure
-
A foundational platform for continued product development and feature expansion