Skip to content

gerry-obrien/GameStateReconstruction_FootballBroadcasts

Repository files navigation

Game State Reconstruction for Football Broadcasts (SoccerNetGS)

This repository contains our Foundations of Deep Learning project on reconstructing football game state from broadcast frames using a two-stage pipeline:

  1. YOLOv8 object detection to localize key entities (players, goalkeepers, referees, ball, other).
  2. ResNet18 crop-based classification to assign detected players/goalkeepers to Left vs Right teams.

🎬 Qualitative results (videos)

Project page with embedded clips (validation + test):
https://gerry-obrien.github.io/GameStateReconstruction_FootballBroadcasts/

YouTube backups (unlisted):

Overlay format:

  • YOLO: <class> <yolo_conf>
  • Players/goalkeepers additionally: TEAM {Left/Right} {team_conf} where team_conf is the ResNet softmax probability for the predicted team.

🧪 Models and evaluation summary

  • YOLOv8: strong detection for large objects (players/referees/goalkeepers); ball detection is the main bottleneck.
  • ResNet18 team classifier: moderate validation accuracy; shows an asymmetric error pattern (more right → left mistakes), which affects team overlays on test footage.

Full quantitative results and plots are reported in the accompanying write-up.

📚 Dataset

We use SoccerNetGS (GameState) frames and labels. Please follow the SoccerNet terms of use and dataset download instructions. This repository does not redistribute the dataset.

👥 Authors

  • G. O'Brien
  • A. Le Saux
  • H. Boisson
  • G. Emmanuel

About

Game State Reconstruction using football broadcasts using YOLO and ResNet18

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors