The sports statistics tool with computer vision algorithms

Software and Hi-Tech

Business need

Every trainer and team owner knows, that the key to success is always players. The better players you have, the higher are chances to win. Therefore the search of the most gifted ones is the main challenge of the business. Team games have lived through two eras.

First, the value of a player was based on their popularity. Of course, the better a player performs, the more popularity they gain. But the more popular player is, the higher is their price. And of course, not all the teams could afford them.

So the other era had started — the era of scouts and sports statistics when the value of a player was estimated not by their popularity, as popularity is a too subtle thing, but by the batch of the statistics parameters. Scouts were spending hours and days, making complex measures and calculations, sifting out gold.

And thus the new era begins when the estimation is done by automatic algorithms, and those scouts succeed, who have the better one.

Solution

Softarex Technologies, Inc. has developed the desktop application that embraced the existing estimation methodologies, formulas, rates, parameters, and furnished it with the cutting-edge computer vision algorithms.

Even though it might sound like a complex tool, it’s easy to use by both techy and non-techy customers due to its simple interface.

Customer

The customers — are agents and trainers, who use a wide range of methodologies and parameters in order to estimate sportsmen technic and rate their value.

Challenge

  • The main goal was to develop a tool that will allow decreasing the amount of manual work done in order to evaluate players’ parameters;
  • The tool was supposed to identify a player on a training record and measure their performance;
  • Apart from providing the instant results, the tool had to export the excel document with all the selected data.

Value Delivered

  • The speed of processing has increased dramatically: now the training videos can be worked-out times faster than they used to;
  • Achieved data is now more precise than the measures gained manually;
  • All the processed footages and selected scenes are stored in an organized way together with a player’s personal info;
  • All the big sports stakeholders can get their benefit from the tool: trainers and players can analyze received results and improve them, while scouts can decry a nugget where everyone other fails.

Project Results

    This desktop application allows to work with sportsmen training recordings and receive all the required parameters automatically, using following features:

  • Video management — a user can upload all necessary videos and add related background info. After the video is uploaded, a user can adjust the playback according to their needs;
  • Scene selection — to save time and resources on processing the whole video, a user can select only the parts he is interested in;
  • Video processing — even though all the features here are listed in order of their appearance in the workflow, this particular one should have gone first as this is the beating heart of the whole application. All the selected parts of the video are automatically processed by the player recognition algorithms. As the result, the application detects the size and position of player’s body parts, required angles and twists. Then it uses its algorithms to processes collected data. And after a minute a user gains the statistics for which he used to struggle for days;
  • Frame selection — after the selected scenes were processed and measures were achieved, a user can surf through frames for the best shots with the most sufficient info on a player’s technic. A user can mark the most useful frames and include them into a table;
  • Results export — after all the achieved results were checked and preferred frames were selected, a user can export the data as a file with all the required info collected into a suitable table.