This project will attempt to visualize a video of a person juggling. I will analyze where the balls travel, and then visualize of their path by drawing arcs over the video showing the paths of the balls and labeling those arc with the siteswap notation of the throw.

For those unfamiliar, siteswap notation is giving a number to each throw, where zero is an empty hand, odd numbers are an increasingly higher throw between hands and even numbers are an increasingly high throw between the same hand.

Using this data, I'll be able to reconstruct an animation of the performance without the video. This could be useful in artistic juggling performances, as I'd be able to project custom imagery onto a performer to enhance the routine.

The end goal of the visualization is something like the above. The arcs drawn will be determined with a quadratic regression, so theyll look clean, and the siteswap numbers will appear on-screen somewhere near the base of the arc. This will update over time as the position of the balls change.

Some stretch goals for the project would be:

1. Match the juggling arcs to some pre-known patterns and display the coloquial name for the pattern.
2. Try to guess the complete siteswap notation, and using that, look up the siteswap on the juggling wikia and display one of their idealized gifs of the pattern alongside the video.
3. Try to detect what the juggler did wrong when they drop a ball (threw to late/early, didnt align balls to a plain, ect).

• Week 5: Basic ball detection and tracking
• Week 6: Trace the balls arcs and try to figure out where an arc starts and stops
• Week 7: Pattern match the balls into basic siteswap throws
• Week 8: Draw arcs and siteswaps back onto the video
• Week 9: Look up siteswaps on juggling wikia
• Week 10: Try to find mistakes by highlighting the first time a ball deviates from the norm

• Siteswap - Wikipedia
• Eye-Balls: Computer Vision in the Circus - Joe Marshall, Mixed Reality Laboratory, University of Nottingham