Event Detection in Football using Graph Convolutional Networks
This addresses the problem of extracting insights from sports data for professional teams and media houses, but it appears incremental as it applies existing GCN methods to football event detection.
The paper tackled automatic event detection in football videos by modeling players and the ball as graphs in each frame, using Graph Convolutional Networks (GCNs) to process unstructured tracking data, and presented results for graph convolutional layers and pooling methods to model temporal context around actions.
The massive growth of data collection in sports has opened numerous avenues for professional teams and media houses to gain insights from this data. The data collected includes per frame player and ball trajectories, and event annotations such as passes, fouls, cards, goals, etc. Graph Convolutional Networks (GCNs) have recently been employed to process this highly unstructured tracking data which can be otherwise difficult to model because of lack of clarity on how to order players in a sequence and how to handle missing objects of interest. In this thesis, we focus on the goal of automatic event detection from football videos. We show how to model the players and the ball in each frame of the video sequence as a graph, and present the results for graph convolutional layers and pooling methods that can be used to model the temporal context present around each action.