16 May Video Analysis
This post was originally published on Rowing by Numbers
The two big issues with the data in the rowing world at the moment is ‘What data actually is it’ but also ‘How do I understand it?’
Back in November I was sat in a launch following a Women’s Trials 8s match. A full championship course side-by-side thriller between two high quality boats. It was great to see two boats go toe-to-toe down the course with the lead changing hands multiple times and a verdict of less than a length. There were lots of positive outcomes from the on water work with the ladies learning a lot about holding onto the red-line for 20 mins both physiologically and mentally.
My role in the following launch was to grab as much video and pictures as possible. This was mainly for the race report (and Instagram of course) but also for some basic video analysis. Ideal ways to give feedback to the coxes on their choice of lines at various situations. Video analysis is known as a great coaching and feedback tool.
This year I have been exploring the Peach Innovations Ltd telemetry system. We were able to fit both of the Women’s 8s involved with the sensors recording gate angles and force at 50Hz. Fortunately we were able to extract the stroke data which provided us with some values for each rower.
In reviewing the telemetry of both boats after the session we were struggling to gain a huge amount of insight when comparing the two boats. Despite the ideal situation of taking place at the same time with the same environmental conditions due to some of the limitations of the Peach software (Powerline) it is quite hard to compare two sets of data (about 1.8 million data points for each boats 20-odd minute piece).
However by exporting some of the raw CSV data I was able to produce some analysis of the two boats. Spreadsheets are common place to most these days but even looking at the processed data, 1000 rows by 30 columns of numbers, it all blends together.
|A screen shot of one of the telemetry raw data spreadsheets|
Data visualisation has come on vastly in the recent years as ‘big data’ has developed so have the ways for displaying not only the data but the insights (see http://www.informationisbeautiful.net/ for some great examples).
Upon reviewing the footage I had taken I had taken it occurred to me it would be great if we could add some of the telemetry information to the video, using the inspiration of a canoe sprint video from Deloitte Denmark:
Some old school (2011) posts from Rowing Musings Blog (a great read) had done something similar with a piece of software called Dashware. Although the blog was a couple years old the principles were a great starting point. The Dashware software was taken over by GoPro recently and is the basis behind their added telemetry overlays, however the original software is available for free.
After experimenting with some of the gauges (a graphical display of information) and creating some of my own the big question was what information did we want to display? A key question these days where ‘big data’ can seem like too much. We wanted to show the power of both boats, we’ve been looking at what a rowers power really means (does it make the boat go faster?).
I had also been working with some of the individual athletes to compare their power output to their HR which would be a really good comparison. We could see how hard people were working, but also see who wasn’t.
|Our data interface for the video overlay|
A bit of excel magic sorted out the data and lined it all up. Unfortunately only one athlete was wearing a HR monitor so all the other athletes HR data was changed using some RAND multiplication in excel. Otherwise the data displayed are actual recorded values.
The crucial bit was being able to compare both the boats (who was going faster) and the athletes in one video. Hence we came up with the above interface, with the individual athletes on each side corresponding to the boat on each side and overall power and rate of each boat in the bottom.
Here’s the final video :
So what did we learn?
We were able to make some conclusions about what the power values output by the telemetry told us about boat speed. If you see both boats move off with about the same speed, with very similar power values, what does that tell you about the values we are getting?
After posting the video we also had one of the athletes come to speak to the coaches concerned how their power values compared to the rest of the squad, the telemetry offers no hiding space in an 8.
Why did we do it?
The answer to this a little blunt, but because we could. It was a great exercise in visualising the data especially to compare two boats in the same situation. But it also gave us some understanding in what we were looking at and the systems validation/accuracy being able to compare against two data sets effectively.
We published it online through the club account and got alot of feedback and engagement from the rowing community which we were interested in the responses of.
Where to go from here?
While this was quite a lot of work and it would be good to do this for some more of the sessions we do to get the video overlay of two crews doesn’t happen a huge amount with full telemetry very often. Imagine the overlay of the Grand Final at HRR? Two mens 8s side by side through the enclosures would be a very interesting comparison.
Linking back to ‘what data is it’ and ‘how do I understand it’ we had some quite in depth discussions to work out how best to display this information but also how to understand it and without the correct visualisation we would not have learnt as much as we could. The technology (free as well) is available but the know how to turn numbers into meaningful decisions and improvements is something that this video overlay of data can provide.