Combining Video With Data In Sports

Neither data nor video analysis should exist in isolation


If a soccer coach examines the biometric data of a 34-year-old centre-back up against their 24-year-old deputy, the results might not flatter the experienced player. Fewer sprints, less distance covered, fewer sudden movements like tackles or lunges to block a cross - the physical data could be taken as a clear indication of the player’s decline in effectiveness on the pitch. The older player is dropped, the youngster takes their place.

If the same soccer coach spends hours sifting through video footage of the games in question they’d see that, from a defender’s point of view, sometimes less is more. Fewer sprints means the defender is in the correct position more regularly and doesn’t need to exert themselves often, less distance covered can mean similar things, and sudden movements like lunges and tackles are last resorts, something the best defenders look to avoid.

In soccer, defense is about communication, positioning, and discipline - only one of these can be measured by wearables. Experience and awareness of the wider game at hand can’t be tangibly measured, and only through observation can coaches get an accurate idea of how useful a player is. The same can be said for that much-lauded quality of ‘vision’. The ability to get your head up and pick the right pass is valued across a host of sports, but again it’s impossible to measure as a numerical metric. Even if a player registers a record number of ‘key passes’, its difficult to discern whether those were the best pass available in that situation, or whether the decision was made quickly enough.

Both biometric data and video analysis are effective ways of measuring performance, but neither can be at their full efficacy in isolation. Only when the two are combined can the true value of both be realized. Data is at its most useful when it’s pulled from where it matters most - the competitive environment, rather than the lab. Add video context to this data and you have a complete picture of what happened on the day. Not all coaches are budding data scientists, far from it, so being able to present the people making the decisions with the most easily digestible and context-focused data is essential.

One company looking to push the boundaries is GoPro. The wearable camera manufacturer has opened its doors to app developers to help improve its product experience through the use of data visualization. At the Wearable Tech in Sport Summit in San Francisco last August, Adam Silver, GoPro’s Director of Strategic Product Partnerships explained the true power of combining video content with data collection. His presentation focused on the ability to overlay metrics onto video content to produce an all-in-one product.

Seeing potential in the idea, GoPro acquired Dashware in spring 2015, a company that overlays data onto video. Having metrics presented in real-time over video content not only gives the data context, it makes the video one of the most effective data visualization tools out there. The company’s GoPro Hero5 Black camera has Dashware’s technology inbuilt, and uses the GoPro’s own ability to collect metrics like speed, distance, vertical, location, etc. to feed it the necessary data.

‘You know your data, you know your device,’ Silver said. ‘You’ve probably spent months, years, however long looking at the data that’s coming from your devices. And you know when something interesting has happened in that particular moment in time. By synchronising that with video, you can pull out that particular video clip and you can drop it into an auto edit, and then you know for sure that you’ve captured the most interesting parts of the video.’

Auto editing is actually one of the key ways in which you can envisage the conflation of data and video having an effect. Through combining the hours of video content collected with data, analytics teams can highlight key moments or moments when a certain action was performed and create a reel. It’s only when the data is synchronized with the video replay that this becomes a possibility, and its something GoPro hopes will change the way coaching teams manage their video analysis.

The next step in this process is to make the data available in real-time, and this is where the technology gets really exciting. Periscope is a developer in the GoPro developer program, for example, and at present they’re exclusively streaming live video. ‘You can imagine a scenario where there’s a telemetry puck that’s connected wirelessly to the camera, and then the telemetry data is streamed with the video over the cellular network,’ Silver said. ‘I want to see that!’ The possibilities for this technology would be endless, and it seems we’re not far from data having a direct effect on sports performance in real time. GoPro hopes to have regular developer conferences to promote further innovation in the area, with designs to become a real player in sports analytics rather than just a hardware manufacturer.

Sport is far from unique in its necessity for contextualised data, but the amount of variables at play on the field, on the court, in the ring, etc. make context even more vital. Once the coaching teams can be presented with not just the data but the corresponding video for context, the use of data in sports will enter a new era.

Vision small

Read next:

Big Data Forecasting In Pharma