Using Eye-tracking to Evaluate Visual Behaviours on GISEnhanced Football Animations

Nian-Hua Liu, J. Félix Salazar, Chuan Chen, Shengkai Wang, Rostislav Nétek, Stanislav Popelka, Liqiu MengAbstracts 2023 · DOI

Abstract

This research explores how GIS-enhanced animations—such as Voronoi diagrams and convex hulls—can facilitate the understanding of complex football tactics. By employing eye-tracking technology to compare the visual search patterns of experts and novices, the study aims to decode the cognitive processes involved in tactical comprehension. This work-in-progress establishes a multi-stage experimental framework to provide empirical evidence on the effectiveness of specialized spatial visualizations in sports analytics.

Introduction

Football is a high-speed, spatio-temporal puzzle. While raw video recordings are the standard for analysis, they are often too "noisy" to clearly convey underlying team structures and tactics, such as Catenaccio or rapid counter-attacks. GIS-enhanced animations—2D abstractions of player movements—have emerged as a solution to lower the "understanding threshold" for viewers. However, empirical evidence regarding how these visualizations affect human cognition is currently lacking. This paper addresses this gap by investigating how different user groups observe and interact with specialized spatial visualizations.

Methodology

The authors propose a multi-stage experimental framework to evaluate visual behavior:

  • Animation Generation: Using Python and the Bokeh library, the researchers created animations that layer GIS concepts onto football play. These include Voronoi Diagrams to visualize a player's "control area" and Convex Hulls to visualize team compactness and defensive pressure.
  • Comparative User Study: Participants are split into two cohorts: experienced football players/fans and inexperienced users.
  • Eye-Tracking Hardware & Software: The study utilizes a Gazepoint eye tracker to capture fixation sequences and Areas of Interest (AOIs). Data is processed using OGAMA (Open Gaze and Mouse Analyzer) and Gazepoint Analysis software.
  • Cognitive Assessment: Pre- and post-experiment questionnaires measure background knowledge and tactical comprehension, alongside short-term recall tasks to evaluate memory of player movements.
Voronoi diagrams visualizing player control areas over time (20s to 80s).

Datasets

The research utilizes custom-generated datasets rather than standard public machine learning repositories:

  • Custom GIS Animations: Created specifically for this study (Liu, 2022) to represent various tactical scenarios.
  • User Metrics: A primary dataset consisting of gaze points, fixation durations, and questionnaire responses from the two participant groups.
Convex hulls visualizing team compactness and defensive pressure over time (20s to 80s).

Key Insights and Ideas

  • The "Expert" Hypothesis: Experienced users are expected to have more efficient visual behaviors, capturing the tactical "story" of the pitch faster with fewer fixations on irrelevant data.
  • GIS as a Cognitive Scaffold: GIS layers act as a bridge, translating raw movement into meaningful spatial information that the human brain can process more intuitively.
  • Visual Analysis Improvement: By identifying expert visual patterns, researchers hope to design better visualization tools that can "train" novices to focus on the right tactical cues at the right time.

Preliminary Results and Future Work

As a work-in-progress presented at ICC 2023, the paper formalizes the methodology for comparing intergroup differences. While full data analysis is ongoing, the successful creation of the GIS-enhanced animations and the experimental design are complete. Future work will expand to more complex tactical scenarios and real-time interactive tools, with the ultimate goal of providing concrete recommendations for improving the design of sports analytics software.

Summary updated

BibTeX
@inproceedings{liu2023using,
  author = {Nian-Hua Liu and J. Félix Salazar and Chuan Chen and Shengkai Wang and Rostislav Nétek and Stanislav Popelka and Liqiu Meng},
  title = {{Using Eye-tracking to Evaluate Visual Behaviours on GISEnhanced Football Animations}},
  booktitle = {Abstracts},
  year = {2023},
  doi = {10.5194/ica-abs-6-144-2023},
}
Figure