MIDS Capstone Project Fall 2017

Sports Summarizer

Sports fans all over the world can't get enough of the sports they love to watch, partly because they do not have enough free time.

One way to help them is to give them an easy option to watch the highlights of the game. However, creating a highlights video is a manual tedious job.

Sports summarizer generates a highlights video from a full length game using machine learning by using a combination of the audio track, scene and image analysis.

The algorithm was trained on soccer games but we believe it can be adapted to any other spectator sport. 

The results are an index file that can be used to play the highlights by skipping to them in the original full length video or by using it to generate a highlights video.

Automatic generation of highlights video is a key component of our vision to eventually create a personalized sports feed customized to the viewer's preferences.

Fans can't watch enough of the sports they love
Fans can't watch enough of the sports they love
Each second's likelihood of being interesting with chosen highlight clips
Each second's likelihood of being interesting with chosen highlight clips

Last updated:

December 19, 2017