Automated detection of physical contact events in youth ice hockey: a player-centric deep learning approach
Authors:
Open Access
- 25 March 2026
- journal article
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Abstract
Youth ice hockey lacks scalable, automated tools to quantify collision exposure and efficiently identify candidate head-impact events from routine game video. This study presents a player-centric, video-based pipeline for detecting physical contact events in single-view full-game footage and demonstrates its utility as a pre-filter to accelerate head-impact dataset creation and support injury surveillance. For training and validation, we constructed labeled player-centric clips around manually annotated contact events; for full-game testing, we applied the same pipeline to continuous game footage. In both settings, the unit of analysis is a fixed-duration (1 s) player-centric clip, and the model outputs one binary label (contact vs. non-contact) per clip. Players were detected using a youth-tuned You Only Look Once (YOLO)v8 model and tracked using StrongSORT with an intersection over union (IoU)-assisted matching cost to improve identity continuity under temporary occlusions. Contact events were manually annotated in 20 youth games (Under-11, Under-13, and Under-15). For each event, annotators recorded the event frame and marked the impacted player (head/body location), which was used to associate the event with the corresponding tracklet (point-in-box). A 60-frame clip spanning a \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
$$\pm 0.5$$ s window around the event (60 fps) was extracted and uniformly subsampled to 30 frames for classification. Non-contact clips were sampled from tracklet windows that did not overlap annotated events, yielding contact: non-contact ratios of \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
$$\approx$$ 1:4, \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
$$\approx$$ 1:6, and \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
$$\approx$$ 1:9. A Temporal Shift Module (TSM) classifier processed each player-centric clip, and we evaluated the effects of crop scale and class imbalance on contact detection performance. The best configuration (15 segments, shift division \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
$$=4$$ , \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
$$1.5\times$$ crop, \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
$$\approx$$ 1:9 training ratio) achieved strong contact detection under evaluation conditions. The final model was then applied to two unseen full-length Under-13 games by partitioning each game into non-overlapping 1 s segments (60 frames at 60 fps), detecting and tracking players, post-processing tracklets, and classifying a player-centric clip for every tracked player. This full-game evaluation performed substantially above a random baseline and provides a practical operating point under realistic class imbalance. As a downstream demonstration, the contact detector served as an effective pre-filter for head-impact review: At the default decision threshold (0.5), 19 of 22 manually identified head impacts occurred in player-centric clips predicted as contact (86.4% head-impact recall), reducing manual review from over three hours to under 30 minutes per game. Overall, the proposed pipeline enables scalable contact-event monitoring in youth hockey and substantially reduces the burden of curating head-impact datasets from full-game video footage.
Keywords
All Related Versions
This publication has 50 references indexed in Scilit:
- A survey of video-based human action recognition in team sportsArtificial Intelligence Review, 2024
- Video analysis of heading and risk of head injury situations in elite international men's football: Does the frequency of headers increase with the level of play?European Journal of Sport Science, 2024
- Social Adaptive Module for Weakly-Supervised Group Activity Recognition2020
- Validation and Comparison of Instrumented Mouthguards for Measuring Head Kinematics and Assessing Brain Deformation in Football ImpactsAnnals of Biomedical Engineering, 2020
- Hidden Two-Stream Convolutional Networks for Action Recognition2019
- Concussive and subconcussive brain trauma: the complexity of impact biomechanics and injury risk in contact sport2018
- Temporal Segment Networks: Towards Good Practices for Deep Action Recognition2016
- Head Impact Biomechanics in Youth Hockey: Comparisons Across Playing Position, Event Types, and Impact LocationsAnnals of Biomedical Engineering, 2011
- The Relationship Between Subconcussive Impacts and Concussion History on Clinical Measures of Neurologic Function in Collegiate Football PlayersAnnals of Biomedical Engineering, 2011
- The Epidemiology of Sport-Related ConcussionClinics in Sports Medicine, 2011