That was a foul! How viewing angles, viewing distances, and visualization methods influence football referees’ decision-making

Christian Vater, U. W. Schnyder, Daniel MüllerGerman 2024 · DOI

Abstract

A football referee must choose the right position on the field in order to perceive a foul accurately. In previous research, decision-making for foul calls was compared across different situations within a match or even across different matches, which questions the comparability of decisions that often depend on the situational context. In this study, we experimentally controlled this situational context to investigate the effect of viewing distance and viewing angle on referees’ decision-making. We recreated real-life decision-making situations using markerless tracking devices and virtual reality (VR) animation processes to render first-person situations. We compared the decision-making accuracy of ten experienced football referees when judging situations from three different viewing angles and distances. Furthermore, we compared their decision-making accuracy in VR scenes with accuracy when viewing the same scenes from a bird’s eye and slow-motion perspective, i.e., with the perspectives used in official referee training. We found that decision accuracy is not affected by viewing angles and distances, but that the referee’s decision certainty is lower when viewing the foul situation from a greater distance. The visualization of the situations had an impact on the referee’s decision-making but only for detecting a simulation (i.e., a player acts like they have been fouled). Slow-motion scenes led to higher decision-making accuracy than VR scenes. Our study suggests a way to recreate VR situations for experimental studies, allowing researchers to control the context of each situation. Our findings help to disentangle the mixed findings of previous studies on distance and viewing angle effects on decision accuracies. Finally, we propose potential avenues to improve VR animation processes in future studies.

Introduction

Football referees must make high-stakes decisions under extreme time pressure. While sports science has long debated the "optimal" position for a referee, previous studies often compared different fouls from different matches, making it impossible to isolate whether a bad call was due to the referee's position or the complexity of the foul itself. This study addresses this lack of experimental control by using Virtual Reality (VR) to "re-play" the exact same foul from multiple distances and angles.

Methodology: Reconstructing the Pitch

The researchers recruited 10 experienced amateur referees and presented them with 18 real-world foul scenarios originally from FIFA and UEFA matches. To isolate the effects of positioning, the study employed a novel two-step approach to recreate these scenes with animated avatars.

The VR reconstruction workflow: from markerless motion-capture to rendered 3D scenes.

Using markerless motion-capture and 3dsMax, the researchers generated 162 variations (18 situations × 3 viewing angles × 3 distances: 5m, 15m, and 25m). Referees judged the situations in three distinct formats to compare traditional training methods with immersive perspectives.

Comparison of visualization methods: (a) bird’s-eye broadcast, (b) slow-motion/zoomed, and (c) animated VR.

Key Insights

The Certainty-Accuracy Gap

Distance (up to 25m) did not significantly lower decision accuracy, but it plummeted the referees' certainty ($p < 0.001$). This suggests that experienced referees are remarkably good at using "advance cues" and situational logic to fill in visual gaps, even when they feel unsure.

Slow-Mo for Technicality, VR for Intent

Slow-motion video led to the highest overall accuracy for technical fouls. However, VR was significantly better for detecting simulations (dives). This implies that the first-person, real-time perspective is crucial for picking up the subtle "deceptive intent" of a player faking a fall, which can be lost in bird's-eye or slowed-down footage.

Debunking the "Natural" Angle

Contrary to some officiating handbooks that suggest a specific "natural" viewing angle (standing between the two players involved), the study found that viewing angle had no significant effect on accuracy or certainty. Referees were also consistently more accurate at identifying actual fouls than they were at correctly identifying "play-on" situations or dives.

Limitations and Future Work

While the study provides a robust framework for experimental control, the authors note that the sample size (10 referees) was small. Future work could improve immersion by using Head-Mounted Displays (HMDs) rather than back-projection screens and by utilizing 3D positional data from real matches to create even more realistic player dynamics. This research paves the way for more sophisticated VR-based training programs that focus on the nuances of player intent and positioning.

Summary updated

Abstract

Abstract A football referee must choose the right position on the field in order to perceive a foul accurately. In previous research, decision-making for foul calls was compared across different situations within a match or even across different matches, which questions the comparability of decisions that often depend on the situational context. In this study, we experimentally controlled this situational context to investigate the effect of viewing distance and viewing angle on referees’ decision-making. We recreated real-life decision-making situations using markerless tracking devices and virtual reality (VR) animation processes to render first-person situations. We compared the decision-making accuracy of ten experienced football referees when judging situations from three different viewing angles and distances. Furthermore, we compared their decision-making accuracy in VR scenes with accuracy when viewing the same scenes from a bird’s eye and slow-motion perspective, i.e., with the perspectives used in official referee training. We found that decision accuracy is not affected by viewing angles and distances, but that the referee’s decision certainty is lower when viewing the foul situation from a greater distance. The visualization of the situations had an impact on the referee’s decision-making but only for detecting a simulation (i.e., a player acts like they have been fouled). Slow-motion scenes led to higher decision-making accuracy than VR scenes. Our study suggests a way to recreate VR situations for experimental studies, allowing researchers to control the context of each situation. Our findings help to disentangle the mixed findings of previous studies on distance and viewing angle effects on decision accuracies. Finally, we propose potential avenues to improve VR animation processes in future studies.
BibTeX
@inproceedings{vater2024that,
  author = {Christian Vater and U. W. Schnyder and Daniel Müller},
  title = {{That was a foul! How viewing angles, viewing distances, and visualization methods influence football referees’ decision-making}},
  booktitle = {German},
  year = {2024},
  doi = {10.1007/s12662-024-00947-5},
}
Figure