##Goals of Juggle Tracker ###Main Goals

• Track juggling balls
• Count catches
• Calculate siteswaps
• Keep track of which hand balls are in

###Other Goals

• Real time juggling analysis
• Background subtraction
• Ball color calibration
  • Works by making circles of regions to get the color of and regions to ignore.
  • Click once to set the center of the circle and then click again to set the radius.
    • Left clicking means you want to get the colors in the circle.
    • Right clicking means to ignore the area in the circle when considering background colors. Primarily used to circle partially covered balls when doing a smart color matching algorithm.
  • Types of color selecting algorithms:
    • Naive: Uses smallest rectangle possible to capture EVERY point circled. Prone to also picking up backgroud of similar color.
    • Smart: Tries to fit the rectangle to include as many circled points as possible and s little background as possible.
• Non-ball props

##Hurdles

• Coming up with a good, general algoritm for identifying catches and throws is difficult
  • Particularly challenging for 1 throws since their nature doesn't give them the arc of other throws
• Backgrounds with similar colors, which will have to be subtracted, but are currently eroded out

##Dependencies

• OpenCV for python (cv2)
• numpy (needed for OpenCV)
• scipy (for k-means clustering done in calibration)

##Demo and Current Status (as ofJuly 23, 2014)

##Authors Kai Smith ([email protected])