Programming language : Python

Libraries used: Pandas, Numpy, Pandas, Scikit-learn, Tensorflow, etc

Database: Firebase (NoSQL)

Tags: time series, classification, machine learning, deep learning, XAI

homepage URL : https://ctrl.golf/

Task

Build golf swing analyzer based on IMU sensor data (time series) to provide personalized feedback instruction to users.

this is a placeholder image
from CTRL Golf

Challenges

  • Noises in sensor data
  • Golf swing analysis (domain knowledge) and interpretation (XAI)
  • Data problems
  • Personalization
  • Limited computing resource of mobile device
this is a placeholder image
from CTRL Golf

Accomplishments

  • Transformed IMU raw data to 3D swing trajectory using pose estimation algorithm
  • Visualized 3D swing trajectory
  • Developed swing performance index (tempo, club angle, club face, etc)
  • Prototyped classification model to predict a direction of ball flight
  • Collected IMU data and tested new version at golf fields