SoccerNet 2026 Player-Centric Ball-Action Spotting:Retraining and Post-Processing Extensions to the FOOTPASS Baselines
2026-06-08 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors developed a system to identify which soccer player is doing which action and when it happens during a game. They improved existing methods by making the model training more efficient, combining different types of information about player actions, and balancing the rare and common actions in the data. They also added steps after the main prediction to clean up the results and used two models together to improve accuracy. Their system performed well on the SoccerNet 2026 challenge tests.
SoccerNetBall-Action SpottingGraph Neural Networks (GNN)Temporal Action Detection (TAAD)Class ImbalanceModel EnsembleGradient CheckpointingMacro F1 ScoreJersey Re-assignment
Authors
Parthsarthi Rawat
Abstract
We describe our system for the SoccerNet 2026 Player-Centric Ball-Action Spotting Challenge, which requires predicting who performs which action and when, across eight classes in broadcast soccer. Building on the three FOOTPASS baselines [1] (TAAD, TAAD+GNN, and TAAD+DST), we contribute four extensions: (1) gradient check pointing to enable full-backbone fine-tuning on a single GPU; (2) fusion of GNN logits into the DST encoder, combining graph-based tactical context with per-player visual features; (3) square-root frequency class weighting to address the 213:1 pass-to-tackle imbalance in the training data; and (4) a post processing pipeline comprising per-class logit gating, temporal frame refinement, jersey re-assignment, and a two-model ensemble. Our system achieves 0.548 Macro F1 on the test set and 0.446 on the challenge set (server evaluation).