Motion Recognition of Assistant Referees in Soccer Games via Selective Color Contrast Revelation

EasyChair Preprint 2604, version 1

Abstract

In this paper, we propose a method to recognize motions of assistant referees in soccer games. Since the major motions of assistant referees in soccer games are closely related to the direction of pointing the flag, positional information of the flag and the upper body can be utilized as an important clue for motion recognition. Based on this observation, we propose to generate and utilize a heatmap image which reveals the contrast between colors of the flag and the upper body. For quantitative evaluation, we used K-league 2017 dataset, and our proposed method achieved 97.56% accuracy surpassing various deep learning-based classifiers. We expect that this study would be used as a useful benchmark for researchers developing sports event recognition systems.

Keyphrases: Heatmap generation, color contrast, motion recognition

@booklet{EasyChair:2604,
  author    = {Yoonhyung Kim and Chanho Jung and Changick Kim},
  title     = {Motion Recognition of Assistant Referees in Soccer Games via Selective Color Contrast Revelation},
  howpublished = {EasyChair Preprint 2604},
  year      = {EasyChair, 2020}}