Stephen Meschke's practice log

View Stephen Meschke's profile, records, practice calendar or everyone's log entries.

View all log users

View older log entries

25th May 2017

Stephen Meschke

# 5 balls, 7 balls, 60 minutes

While training 5b overheads, I was hit in the eye with a ball. My eye hurt and watered a lot. I had to stop juggling for several minutes. It was pretty frustrating, but I did some jumprope and then moved on to 7b.

Ran 7b for 100 catches (to collect) on the fourth or 5th try. Working on 7b endurance has really helped my control with 7b.

Total practice time: 60 minutes

Location: Gym

Comments (0)

20th May 2017

Stephen Meschke

# 6 balls, 4 clubs, 35 minutes

5b. cascade
5b: Transition from (6,6)(6x,2)* to (6x, 4)* and back.
6b: Transition from sync to async and back, 9555
4c: Transition from sync to async and back, 534

Total practice time: 35 minutes

Location: Park

Comments (0)

17th May 2017

Stephen Meschke

# vision research, 1 minutes

Collected data for the patterns 72727: https://drive.google.com/open?id=0B7QqDexrxSfwYWFyamRwZlY1bFE

So ugly! Some of these patterns look really pretty, like 432 and 645, and some look really ugly, like 72727 and 50505.

I think that the patterns that are really pretty have good syncopation, and the patterns that are ugly are arrhythmic. When I juggle 'snake' type patterns (50505, 72727, etc...), I tend to cheat the timing to make the pattern easier. Next, I will collect data on the pattern 7070707 to test this hypothesis.

Still need to work on writing a small-talk post to present the data set.

# 5 balls, 7 balls, 25 minutes

Didn't sleep well last night, and didn't really feel well today. Juggle anyway, why not?

I have been working on overheads a lot lately, and tried 5b overheads today. Some good runs, around 11 catches (I'll call that a PR and log in records).

After juggling a lot of 7b cascade the last two days, my arms were tired. After a few runs of 7b cascade I rage quit and threw the balls like an angry child. Probably should have skipped 7b. all together. Need more optimal training structure.

Total practice time: 26 minutes

Locations: Linux, Gym

Comments (0)

16th May 2017

Stephen Meschke

# 7 balls, 5 clubs, 65 minutes

Tried to get a 7 ball PR, failed @ 180ish. Some runs felt super solid though. Thinking about going back to russian style balls for endurance.

Did a 5 club drill where I started out by throwing one triple and then continuing in doubles. When doing this drill it is necessary to catch the first club before letting go of the fifth club.

# vision research, 1 minutes

Still working on data set. I did two new patterns; 645 and a 5 ball half shower: https://drive.google.com/drive/folders/0B7QqDexrxSfwcmROb1ByNkhqOEU?usp=sharing

When looking at the graph for the half shower* I found something very interseting. There is a point where four lines intersect. I didn't expect to see that. I postulate that all sync patterns have a quadruple point.


*Link to graph for half shower: https://drive.google.com/file/d/0B7QqDexrxSfwaE5NQmtrVjJPVVE/view?usp=sharing

Total practice time: 66 minutes

Locations: Gym, Linux

Comments (2)

12th May 2017

Stephen Meschke

# vision research, 1 minutes

Working on creating a juggling dataset.

Creating a juggling dataset was my original goal, but it kinda morphed into this project: https://www.reddit.com/r/juggling/comments/6aa9ov/juggling_pattern_waveforms/

There was some interest in the data that I used to create the graphs on the video, which instilled hope in me that there is some interest in a juggling data set.

What is a juggling data set anyway?

* The data set consists of several clips of juggling (about 600 frames long each)
* Each clip of juggling has an associated .csv file that contains the locations of each ball in each frame.
* The data gets 'big' pretty fast. For a 10 second clip with 5 balls there are 5*10s*60fps = 3000 observations

Alongside juggling, I have also been training computer vision. I have created/learned to use some programs that can, with a little help, track the locations of the balls. Under optimal conditions, I can track with 100% accuracy fully automated. Under good conditions, I can track at 98% accuracy with 95% automation.

Optimizing the conditions before filming juggling is not difficult; have good lighting and a high degree of contrast between the balls and the other objects in the frame. It helps to use different brightly colored balls too.

Tracking 5 balls for 600 frames took me about 10 minutes. With this level of automation, it has become practical to start building a data set.

Here is a link to the data that I have created so far: https://drive.google.com/drive/folders/0B7QqDexrxSfwcmROb1ByNkhqOEU?usp=sharing

I would like to introduce this idea with a post in small talk, but I am not sure how to go about that. I estimate that less than 50% of jugglingedge.com users are really into data science.

Total practice time: 1 minutes

Location: Linux

Comments (2)

28th March 2017

Stephen Meschke

# vision research, 1 minutes

Making a cascade circular:

https://youtu.be/S-955wQ7FGw

Total practice time: 1 minutes

Location: Linux

Comments (0)

17th March 2017

Stephen Meschke

# vision research, 1 minutes

Found this: https://en.wikipedia.org/wiki/Two-streams_hypothesis#Two_visual_systems

The two-streams theory says that there is an identification stream and a motion stream.

One visual stream goes to the Ventral System. This is the system that identifies the object. The Ventral System works slowly and identifies the object in greater resolution. When the Ventral System is active, the head is moving and the object is fixed.

The other visual stream goes to the Dorsal System. This system identifies motion. The Dorsal system works quickly and stores information for only milliseconds. When the Dorsal System is active, the head is fixed and the object(s) are moving.

Consider examples:

You are hungry and you come across some wild berries. You have to examine the wild berries and access whether they are edible. System used: Ventral System.

You are resting in the forest, and you see a spider descend from a tree. You need to move so that the spider doesn't land on you. System used: Dorsal System.

My experiences with juggling support the two-streams theory.

Total practice time: 1 minutes

Location: Linux

Comments (0)

14th March 2017

Stephen Meschke

# vision research, 1 minutes

progress on tracking program using stereo cameras:

https://youtu.be/ZvqqKx7sb5Q

Does anyone here know math?

I can't figure out how the length of the line in the video relates to the distance the ball is from the cameras.

Total practice time: 1 minutes

Location: Linux

Comments (1)

7th March 2017

Stephen Meschke

# 7 balls, 5 clubs, 1 minutes

ran 7b for 120+ catches, ran 5c for 120+catches

# vision research, 1 minutes

Lots of computer vision researchers base their concepts on biological systems.

I decided to make a video that helps explain how I see for juggling:

https://youtu.be/GQiKjZaoMwU

Lot's of it doesn't make sense, but it works. I am having some problems with object tracking.

Total practice time: 2 minutes

Locations: Gym, Linux

Comments (1)

28th February 2017

Stephen Meschke

# vision research, 1 minutes

This video talks about human stereo vision: https://youtu.be/gGRPHfQhawI

It's from a course in computer vision.

The point of the video is that binocular vision relies on a low level process that directly fuses the two images.

Applied to juggling; this means that a ball that is far away is determined to be far away because in the fused image from both eyes it appears as a single ball. A ball that is nearer the eyes appears near because in the fused image two overlapping (occluded) circles are seen.

In most, if not all aspects, the capabilities of computer vision on desktop computers is greater than that of human vision. With the exception of determining depth from stereo vision.

Computer vision sucks at finding depth. Human vision is great at determining which objects are near and far.

playlist: https://youtu.be/gGRPHfQhawI?list=PLAwxTw4SYaPnbDacyrK_kB_RUkuxQBlCm

Total practice time: 1 minutes

Location: Linux

Comments (0)

View older log entries

Tags

Multiple tag selection widget

>5 endurance, 3-5 Clubs, 3-6 balls, 3-6 clubs, 3-7 balls, 4 clubs, 5, 5 balls, 5 clubs, 5-6 balls, 5-7 balls, 5-8 balls, 6 balls, 7 balls, balance, balance spinner, dwell time, handstand, Joggling, Juggling Simulator, PR, Rola-bola, routine, runfinder, training, tutorial, vision research