Contrastive Learning for Sports Video: Unsupervised Player Classification

TL;DR

This paper introduces an unsupervised contrastive learning method for classifying players by team in sports videos, achieving high accuracy with minimal data and enabling real-time analysis without prior jersey information.

Contribution

It presents a novel contrastive learning approach for unsupervised team classification in sports videos, outperforming previous methods and functioning effectively with limited frames.

Findings

94% accuracy after training on a single frame

97% accuracy within 500 frames (~17 seconds)

Enables real-time team-conditional heat maps

Abstract

We address the problem of unsupervised classification of players in a team sport according to their team affiliation, when jersey colours and design are not known a priori. We adopt a contrastive learning approach in which an embedding network learns to maximize the distance between representations of players on different teams relative to players on the same team, in a purely unsupervised fashion, without any labelled data. We evaluate the approach using a new hockey dataset and find that it outperforms prior unsupervised approaches by a substantial margin, particularly for real-time application when only a small number of frames are available for unsupervised learning before team assignments must be made. Remarkably, we show that our contrastive method achieves 94% accuracy after unsupervised training on only a single frame, with accuracy rising to 97% within 500 frames (17 seconds of game time). We further demonstrate how accurate team classification allows accurate team-conditional heat maps of player positioning to be computed.