Design for good.
ezgif.com-video-to-gif (2).gif

SwingGo

Using Computer Vision to improve Golf Swing

Logo1.png
 

Using computer vision to improve your Golf swing

 
 

ROLE DURATION TOOLS TYPE

UI/UX Designer 3 Months Figma, Sketch UX Design


Point of View

Athletic ability generally allows you to excel at most sports but this isn’t the exact case with golf. Golf isn’t about being the quickest on your feet, being the strongest, nor is it having the most endurance--its about skill and calculated techniques.

Learning how to play golf can be quite difficult and expensive (hence, the common phrase “a rich man’s sport”) but finding a compatible coach doesn’t have to be. One of the biggest challenges that inhibits Golfers from improving their swing is understanding and identifying areas that they need to work on. Using computer vision, SwingGo acts as a convenient and inexpensive virtual coach by providing immediate and personalized vocal feedback based on the data points of your swing which means the faster you receive feedback, the faster you can start practicing efficiently!


Core Features

Screen+Shot+2019-05-11+at+4.29.20+PM.jpg
  1. REAL-TIME FEEDBACK

    The instantaneous vocal feature mimics the experience of receiving personable and direct advice from a coach while the textual transcription allows users to process the information.

  2. MOTION DETECTION

    With pose summation, the front facing camera of your device detects, tracks, and analyzes the angles and movement of your swing to ensure accurate feedback.

  3. PROGRESS TRACKER:

    Once your data accumulates, SwingGo visualizes your growth over time and estimates your Handicap, a numerical measure of a Golfer’s potential ability.

Want to give it a swing? Play with the application.


Storyboards

Three different design ideas that address the user needs exemplified by the Point of View

  1. Overcoming financial barriers 2. Identifying Mistakes 3. Finding compatible coach

storyboards.png

Rapid Prototyping

1. Prototype A (Instantiates with Storyboard 3)

By game-ifying the process, Swingo makes the experience of learning how to play and building the foundations of golf more accessible. Not only is it a fun tool but it's also crucial to establish a strong foundation before diving into harder skills. Golf can be difficult to learn but Swingo works because users are motivated to learn because they can level up, talk to their friends, upgrade their avatar by earning points, visualize their growth through analytics, and look at their achievements.   

2. Prototype B (Instantiates with Storyboard 1 & 2)

This design not only helps users recognize what they can improve upon but more importantly, how they can improve by providing constructive criticism, video support, and written instructions based on the user’s video submission. The archive setting stores each video session, allowing users to go back in time.

Although the social element of game-ifying the process of learning was creative, we ultimately pursued the concept behind Prototype B after conducting user interviews because users were more interest in a straight to the point process that would allow them to improve quickly.


Heuristic Evaluations

Evaluators: Jamie Shi, Angela Zhu, and Kylee Peng

Key takeaways:

  • Scrap Archive concept and replace with visual summary of progress in the Profile view so that users do not have to specifically remember their performance over the span of days, weeks, months

  • Homepage should only consist of the camera feature and navigation menu

  • Include ability to “Save” and “Delete” footage


Wireframes

view  prototype

User testing

UserTest.png

Observations

  • Testers were receptive and interacted with positive voice feedback (thumbs up, saying “thank you” or smiling)

  • Negative reactions (frustration, annoyance, attitude) to negative feedback and wait time to load

  • Testers were startled and surprised when opening the application as soon as the camera screen launched

  • Testers could not find the Log Out Button on the profile page

Proposed Changes

  • Change placement of Log Out Button for discoverability

  • Add brief description to Login page to demonstrate what our app entails to reduce shock

  • Add different content for voice feedback and reduce length so Testers can process information

  • Provide graphic instructions on the camera page (“Position within frame” and “Stand parallel”)

  • Add textual feedback on screen to supplement voice feedback

  • Ability to turn on/off voice feedback


A/B Testing (N=36)

We created a new design (Design B) where the log out button would be easier to find since the testers did not realize that scrolling was required to log out of the application with Design A which was a task that all testers failed to complete during user testing. To measure which design is better, we calculated how many testers were able find and therefore click the log out button. Based on Google Analytics and Chi-Square Testing, we decided to move forward with Design B.

Design B’s (profileAlt) probability of outperforming Design A (Original) is 80.6%.

Design B’s (profileAlt) probability of outperforming Design A (Original) is 80.6%.

Design A (Profile) had almost twice as much page views compared to Design B. We are speculating that testers went back and forth on the page when they couldn’t find the logout button. Meanwhile, with Design B the log out button is easily discoverable so testers did not switch between pages as much.

Design A (Profile) had almost twice as much page views compared to Design B. We are speculating that testers went back and forth on the page when they couldn’t find the logout button. Meanwhile, with Design B the log out button is easily discoverable so testers did not switch between pages as much.

The chi-square statistic is 3.8961 and the p-value is 0.048398. The result is significant at p < 0.05. Since P value (0.048 < 0.5), the result is significant and we reject the null hypothesis and conclude that there a significant difference exists between Design A and B. We are speculating that the relocation of the log out button made it more discoverable/visible which is why there were more clicks in comparison to “didn’t click”.

The chi-square statistic is 3.8961 and the p-value is 0.048398. The result is significant at p < 0.05. Since P value (0.048 < 0.5), the result is significant and we reject the null hypothesis and conclude that there a significant difference exists between Design A and B. We are speculating that the relocation of the log out button made it more discoverable/visible which is why there were more clicks in comparison to “didn’t click”.


Pitch and Poster

SwingGoPoster.png

Teammates: Broderick Higby and Martin Magsombol