PySport - Patrick Lucey - Generating complete tracking data using event data and remote video

Описание: PySport Meetup London - March 2024

Resources: https://github.com/hhughes-dot/Scalin...

PySport - Patrick Lucey - Generating complete tracking data using event data and remote video

Optical player and ball tracking data has been around for 25 years. In that time, it has evolved from manual to automatic collection using in-venue cameras and more recently from remote video. There still exists a sizable difference between tracking data collected from in-venue compared to remote video, due to the various occlusions that occur during a broadcast (close-ups, replays etc.). Using the multimodality of football data (i.e., event data and player positions), we are able to use Generative AI techniques to generate complete tracking data merged with event data.

In this talk, Patrick will first highlight how this is done and the implications of this work. Then he will showcase an example game using a tracking and event data visualizer that Stats Perform has created in a Jupyter notebook that will be released as part of tonight’s meetup. Having a common tool in Python, which enables practitioners to easily visualize, search and scrub through a game via time or event, can help get people started with these datasets quickly. As things progress, Patrick is aiming to add more of these tools to easily visualize and consume outputs such as passing options and shape as well as other enriched outputs which he will cover in his talk.

Patrick is the Chief Scientist at Stats Perform. He has studied and worked in the AI field for the past 20 years, holding research positions at Disney Research and the Robotics Institute at Carnegie Mellon University as well as spending time at IBM’s T.J. Watson Research Center while pursuing his PhD. Patrick hails from Australia, where he received his BEng(EE) from the University of Southern Queensland and his doctorate from Queensland University of Technology. He has authored more than 100 peer-reviewed papers and has been a co-author on papers in the MIT Sloan Best Research Paper Track, winning best paper in 2016 and runner-up in 2017 and 2018.