Classification of Australian football kick types in-situation via ankle-mounted inertial measurement units.

The utility of inertial measurement units (IMUs) for sporting skill and performance analysis during training and competition is advantageous for enhancing the objectivity of athlete monitoring. This study aimed to classify Australian Rules football (AF) kick types in an applied environment using ankle-mounted IMUs. IMUs and video capture of a controlled protocol, including four kick types at varying distances, were recorded during a single testing session with female AF athletes (n = 20). Processed IMU data were modelled using support vector machine classifier, random forest, and k-nearest neighbour algorithms under a 2-Kick, 4-Kick, and kick distance (10, 20, 30 m) conditions. The random forest model showed the highest results for overall classification accuracy (83% 2-Kick and 80% 4-Kick), test F1-score (0.76 2-Kick and 0.81 4-Kick), and AUC score (0.58 2-Kick and 0.60 4-Kick). Kick distance classification showed a model test and class weighted F1-score of 0.63 and overall accuracy of 64%, respectively. This study highlights the potential for an applied semi-automated AF training kick detection and type classification system using IMUs.

The relationship of team and individual athlete performances on match quarter outcome in elite women's Australian Rules football.

OBJECTIVES: To evaluate the relationships between the athlete distribution of team performance indicators and quarter outcome in elite women's Australian Rules football matches. DESIGN: Retrospective longitudinal cohort analysis. METHODS: Thirteen performance indicators were obtained from 56 matches across the 2017 and 2018 Australian Football League Women's (AFLW) seasons. Absolute and relative values of 13 performance indicators were obtained for each athlete, in each quarter of all matches. Eleven features were further extracted for each performance indicator, resulting in a total of 169 features. Generalised estimating equations (GEE) and regression decision trees were run across the different feature sets and dependent variables, resulting in 22 separate models. RESULTS: The GEE algorithm produced slightly lower mean absolute errors across all dependent variables and feature sets comparative to the regression decision tree models. Quarter outcome was more accurately explained when considered as total points scored comparative to quarter score margin. Team differential and the 75th percentile of individual athlete Inside 50s were the strongest features included in the models. CONCLUSIONS: Modelling performance statistics by quarter outcomes provides specific practical information for in-game tactics and coaching in relation to athlete performances each quarter. Within the current elite women's Australian Rules football competition, key high performing individual athletes' skilled performances within matches contribute more to success rather than a collective team effort.

Machine and deep learning for sport-specific movement recognition: a systematic review of model development and performance.

Objective assessment of an athlete's performance is of importance in elite sports to facilitate detailed analysis. The implementation of automated detection and recognition of sport-specific movements overcomes the limitations associated with manual performance analysis methods. The object of this study was to systematically review the literature on machine and deep learning for sport-specific movement recognition using inertial measurement unit (IMU) and, or computer vision data inputs. A search of multiple databases was undertaken. Included studies must have investigated a sport-specific movement and analysed via machine or deep learning methods for model development. A total of 52 studies met the inclusion and exclusion criteria. Data pre-processing, processing, model development and evaluation methods varied across the studies. Model development for movement recognition were predominantly undertaken using supervised classification approaches. A kernel form of the Support Vector Machine algorithm was used in 53% of IMU and 50% of vision-based studies. Twelve studies used a deep learning method as a form of Convolutional Neural Network algorithm and one study also adopted a Long Short Term Memory architecture in their model. The adaptation of experimental set-up, data pre-processing, and model development methods are best considered in relation to the characteristics of the targeted sports movement(s).

Analysis of Training Loads in Elite Under 18 Australian Rule Football Players.

Cust, EE, Elsworthy, N, and Robertson, S. Analysis of training loads in elite under 18 Australian rule football players. J Strength Cond Res 32(9): 2521-2528, 2017-Differences in training loads (TLs) between under 18 (U18) Australian rules football (AF) state academy-selected and state academy-nonselected players were investigated. Players were categorized relating to their highest representative level: state academy-selected (n = 9) and TAC cup-level players (n = 38). Data were obtained from an online training-monitoring tool implemented to collect player training and match information across a 20-week period during the regular season. Parameters modeled included AF skills, strength, and other sport training sessions. Descriptive statistics (mean ± SD) and between-group comparisons (Cohen's d) were computed. A J48 decision tree modeled which TL variables could predict selection level. Pooled data showed 60% of weekly training duration consisted of AF training sessions. Similar AF TL were reported between state academy and TAC cup players (1,578 ± 1,264 arbitrary units (AU) vs. 1,368 ± 872 AU; d = 0.05). Although higher TLs were reported for state-selected players comparative with TAC cup in total training (d = 0.20), core stability (d = 0.36), flexibility (d = 0.44), on-feet conditioning (d = 0.26), and off-feet conditioning (d = 0.26). Decision tree analysis showed core stability duration and flexibility TL, the most influential parameters in classifying group selection (97.7% accuracy TAC cup level; 35.8% accuracy state academy level). Insights of U18 AF players' weekly training structures, loads, and characteristics of higher achieving players are provided. This study supports the application of training diaries and session rating of perceived exertion for TL monitoring in junior athletes.