Club fitting is an important part of the Titleist customer journey and often the first opportunity to build a data-driven relationship. The key to this data-driven relationship is that good, clean data is collected from customers at the time of fitting. Titleist’s professional club fitters, while excellent at finding the right equipment for their customers at the time of purchase, do not have a high level of accuracy inputting data about customer equipment. Titleist needed a solution that would automate the process of gathering club data using computer vision so that the club fitters could concentrate on what they do best – serving the needs of the customers in front of them.
The deep learning-based OCR solution the Dynam.AI team built for Titleist goes a step beyond standard computer vision and enables work on “unseen” objects. Unlike most computer vision problems, where the objects can be differentiated by shape alone, our problem for Titleist was to identify the exact model of a golf club head and shaft. Different models of golf club heads and shafts are very similar when looking at the shape alone. This forced us to come up with a solution that is similar to how we, as humans, would recognize the golf club head and shaft model. Our model identified the golf club head independent of the shaft as a driver, iron, or wedge. Next we trained the solution to find the location of the marks on the club head that distinguish the different models. Once the solution could locate the distinctive markers, it was able to identify the markers and distinguish different club heads. The final product was an easy to use iOS app that the professional club fitters could easily take with them to club fittings, enabling them to do the necessary data entry by simply pointing the device camera at the customer’s golf clubs.