Classification of kinematic swimming data with emphasis on resource consumption

Jensen, Ulf; Prade, Franziska; Eskofier, Bjoern M.

doi:10.1109/bsn.2013.6575501

Cited by 28 publications

(30 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A dominant approach was the classification of main characterising actions for each sport. For example, serve, forehand, backhand strokes in tennis (Connaghan et al, 2011;Kos & Kramberger, 2017;Ó Conaire et al, 2010;Shah, Chokalingam, Paluri, & Pradeep, 2007;Srivastava et al, 2015), and the four competition strokes in swimming (Jensen, Blank, Kugler, & Eskofier, 2016;Jensen, Prade, & Eskofier, 2013;Liao et al, 2003;Victor et al, 2017). Several studies further classified sub-categories of actions.…”

Section: Experimental Designmentioning

confidence: 99%

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

Cust

Sweeting

Ball

et al. 2018

Journal of Sports Sciences

219

148

View full text Add to dashboard Cite

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 sportspecific movement and analysed via machine or deep learning methods for model development. A total of 52 studies met the inclusion and exclusion criteria. Data preprocessing, 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 setup , data pre-processing, and model development methods are best considered in relation to the characteristics of the targeted sports movement(s).

show abstract

Section: Experimental Designmentioning

confidence: 99%

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

Cust

Sweeting

Ball

et al. 2018

Journal of Sports Sciences

219

148

View full text Add to dashboard Cite

show abstract

“…These constraints prompt behavioral adaptations, such as (i) increasing SR to overcome the drag created by the swimmer in the next lane, (ii) regulating swimming speed when approaching the wall to perform a turn, or (iii) showing fatigue at the end of a race (Dadashi et al, 2016). Also, the acceleration data recorded over a training session can help coaches to distinguish between swimming styles (Pansiot et al, 2010; Hou, 2012; Jensen et al, 2013; Ohgi et al, 2014; Mooney et al, 2015b), but not between two different signals emerging from the same swimming style. This means that a single sensor positioned on the chest can distinguish the signals obtained in freestyle and backstroke from those obtained in butterfly and breaststroke through the general shape of the acceleration versus time curves (Le Sage et al, 2011; Jensen et al, 2013; Ohgi et al, 2014) (please refer to Le Sage et al, 2011; Ohgi et al, 2014, for a depiction of these curves).…”

Section: Investigation Of High-order Parameters To Characterize Coordmentioning

confidence: 99%

“…Also, the acceleration data recorded over a training session can help coaches to distinguish between swimming styles (Pansiot et al, 2010; Hou, 2012; Jensen et al, 2013; Ohgi et al, 2014; Mooney et al, 2015b), but not between two different signals emerging from the same swimming style. This means that a single sensor positioned on the chest can distinguish the signals obtained in freestyle and backstroke from those obtained in butterfly and breaststroke through the general shape of the acceleration versus time curves (Le Sage et al, 2011; Jensen et al, 2013; Ohgi et al, 2014) (please refer to Le Sage et al, 2011; Ohgi et al, 2014, for a depiction of these curves). It is therefore possible for coaches and swimmers to accurately record the swimming distances in each style during a training session and adapt exercise (e.g., swimming at high SR, with paddles or fins, or against a high fluid flow) in preparation for future competitions.…”

Section: Investigation Of High-order Parameters To Characterize Coordmentioning

confidence: 99%

Behavioral Dynamics in Swimming: The Appropriate Use of Inertial Measurement Units

Guignard¹,

Rouard

Chollet

et al. 2017

Front. Psychol.

View full text Add to dashboard Cite

Motor control in swimming can be analyzed using low- and high-order parameters of behavior. Low-order parameters generally refer to the superficial aspects of movement (i.e., position, velocity, acceleration), whereas high-order parameters capture the dynamics of movement coordination. To assess human aquatic behavior, both types have usually been investigated with multi-camera systems, as they offer high three-dimensional spatial accuracy. Research in ecological dynamics has shown that movement system variability can be viewed as a functional property of skilled performers, helping them adapt their movements to the surrounding constraints. Yet to determine the variability of swimming behavior, a large number of stroke cycles (i.e., inter-cyclic variability) has to be analyzed, which is impossible with camera-based systems as they simply record behaviors over restricted volumes of water. Inertial measurement units (IMUs) were designed to explore the parameters and variability of coordination dynamics. These light, transportable and easy-to-use devices offer new perspectives for swimming research because they can record low- to high-order behavioral parameters over long periods. We first review how the low-order behavioral parameters (i.e., speed, stroke length, stroke rate) of human aquatic locomotion and their variability can be assessed using IMUs. We then review the way high-order parameters are assessed and the adaptive role of movement and coordination variability in swimming. We give special focus to the circumstances in which determining the variability between stroke cycles provides insight into how behavior oscillates between stable and flexible states to functionally respond to environmental and task constraints. The last section of the review is dedicated to practical recommendations for coaches on using IMUs to monitor swimming performance. We therefore highlight the need for rigor in dealing with these sensors appropriately in water. We explain the fundamental and mandatory steps to follow for accurate results with IMUs, from data acquisition (e.g., waterproofing procedures) to interpretation (e.g., drift correction).

show abstract

“…The problem we have assigned in the evaluation experiment is classifying butterfly and breaststroke from one another using the inertia sensor data acquired during a swimming race. The swimming style classifier constructed in many earlier studies have commonly confused butterfly and breaststroke [17][18][19][20]. This is attributable to the fact that butterfly and breaststroke are more similar than other swimming motions.…”

Section: Experimental Purpose and Outlinementioning

confidence: 99%

“…, a 42 }), as shown in Table 1. These are the same features used in earlier studies on swimming style classification [17][18][19][20].…”

Section: Conversion Into Featuresmentioning

confidence: 99%

Feature Selection Algorithm Considering Trial and Individual Differences for Machine Learning of Human Activity Recognition

Omae¹,

Takahashi²

2017

JACIII

View full text Add to dashboard Cite

In recent years, many studies have been performed on the automatic classification of human body motions based on inertia sensor data using a combination of inertia sensors and machine learning; training data is necessary where sensor data and human body motions correspond to one another. It can be difficult to conduct experiments involving a large number of subjects over an extended time period, because of concern for the fatigue or injury of subjects. Many studies, therefore, allow a small number of subjects to perform repeated body motions subject to classification, to acquire data on which to build training data. Any classifiers constructed using such training data will have some problems associated with generalization errors caused by individual and trial differences. In order to suppress such generalization errors, feature spaces must be obtained that are less likely to generate generalization errors due to individual and trial differences. To obtain such feature spaces, we require indices to evaluate the likelihood of the feature spaces generating generalization errors due to individual and trial errors. This paper, therefore, aims to devise such evaluation indices from the perspectives. The evaluation indices we propose in this paper can be obtained by first constructing acquired data probability distributions that represent individual and trial differences, and then using such probability distributions to calculate any risks of generating generalization errors. We have verified the effectiveness of the proposed evaluation method by applying it to sensor data for butterfly and breaststroke swimming. For the purpose of comparison, we have also applied a few available existing evaluation methods. We have constructed classifiers for butterfly and breaststroke swimming by applying a support vector machine to the feature spaces obtained by the proposed and existing methods. Based on the accuracy verification we conducted with test data, we found that the proposed method produced significantly higher F-measure than the existing methods. This proves that the use of the proposed evaluation indices enables us to obtain a feature space that is less likely to generate generalization errors due to individual and trial differences.

show abstract

Classification of kinematic swimming data with emphasis on resource consumption

Cited by 28 publications

References 11 publications

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

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

Behavioral Dynamics in Swimming: The Appropriate Use of Inertial Measurement Units

Feature Selection Algorithm Considering Trial and Individual Differences for Machine Learning of Human Activity Recognition

Contact Info

Product

Resources

About