Conceived, designed and implemented a computer vision algorithm to detect and track golf balls flying downrange up to approximately 50 yards from an iPhone 5 video recording. Reprojected from 2-dimensional video into 3-dimensions, and extrapolated trajectory using Newtonian mechanics and Magnus effect. Achieved performance of 8% average error in tracking ball flight distance compared with gold standard on a sample of swings averaging 150 yards.

All algorithms and software were written in high-performance, cross platform C++ code that runs on iOS, Android, Linux, OS X, and Windows.


  • $2500 grant from Yale Entrepreneurship Institute VCP


ProTrack Technology Demonstration
Chart comparing distance measurements of the ProTrack versus those of the TrackMan gold standard. As can be seen, the relationship is quite linear. Mean error was 12.5 yards.


C++, Python, OpenCV, CMake, Git

More Information

Before building a smartphone-based golf ball flight tracker, the plan was to build an inexpensive radar to serve the same purpose. Here is the original slide deck for that idea.