Online Monitoring of Swimmer Training Using a 3D Accelerometer

Identifying Swimming and Swimming Style

Marko Topalovic

1,2

, Simon Eyers

, Vasileios Exadaktylos

, Jan Olbrecht

, Daniel Berckmans

and Jean-Marie Aerts

Division Measure, Model & Manage Bioresponses (M3 BIORES), Department of Biosystems,

KU Leuven, Leuven, Belgium

Respiratory Division, University Hospital Leuven, Department of Clinical and Experimental Medicine,

KU Leuven, Leuven, Belgium

Department of Biosystems, Faculty of Bioscience Engineering, KU Leuven, Leuven, Belgium

Keywords: Swimming, Accelerometer, Stroke Recognition, Activity Monitoring.

Abstract: In the process of optimizing training efficiency and improving results of the athletes, technology has

increasing share. Wearable sensors, especially those measuring motion are lately acquiring more and more

interest. In this paper, we aimed to develop online monitoring tool of swimming training, more in particular

algorithm for detection of swimming and turning events using 3D accelerometer. Additionally, algorithm

should be able to discriminate between performed swimming styles. This study included data of 10

swimmers who swam on predefined protocol for 1200m. Each swimmer was equipped with wireless

waterproof 3D accelerometer attached over right wrist. Algorithm showed high accuracy of 100% for

detection of swimming and turning activity. Additionally, detection of swimming styles such as crawl,

breaststroke and backstroke resulted of 100% true positive rate. However, true positive rate decreased to

95% for detection of butterfly event. To conclude, we demonstrate that swimming activity together with

style recognition can be registered using wireless waterproof 3D accelerometer. Furthermore, we show that

such detection can be automatized and performed in an online mode. Taken together, this development leads

to a useful online monitoring tool of swimming training.

1 INTRODUCTION

With the recent boost in technology development,

we are witnessing greater involvement of new

technical inventions in the competitions, as well as

in the training process of athletes. In the modern

sport, it is beneficial not only to test athletes in the

laboratory environment, but also to give appropriate

guidance of the training process (Maglischo 1993).

The latter is essential as it will optimize training

efficiency and thus improve results that the athlete

can obtain.

The application of accelerometers in sports is

increasing either by measuring activity levels using

mobile phones (Shumei et al. 2010), characterizing

biomechanical activity (Auvinet et al. 2002), or

developing specialized accelerometer-based devices

to measure energy expenditure (Wixted et al. 2007).

However, the use of accelerometers in

swimming, training or competition, is not yet

widespread. Most of the work is based on algorithms

with complex features to discriminate between

limited swimming styles (Siirtola et al. 2011), with

algorithms that work only in offline mode (Jensen et

al. 2013) or on algorithms with interesting concepts

that are yet to be validated in the training process

and larger number of subjects (Bachlin and Troster

2012; Le Sage et al. 2010).

Therefore, we aimed to develop an online

monitoring tool to follow the swimming training

using a three dimensional (3D) accelerometer

attached on a swimmer’s wrist. Moreover,

developed algorithm should be able to detect

whether the subject swims or turns, as well as which

of the four competitions swimming styles is being

used.

111

Topalovic M., Eyers S., Exadaktylos V., Olbrecht J., Berckmans D. and Aerts J..

Online Monitoring of Swimmer Training Using a 3D Accelerometer - Identifying Swimming and Swimming Style.

DOI: 10.5220/0005134501110115

In Proceedings of the 2nd International Congress on Sports Sciences Research and Technology Support (icSPORTS-2014), pages 111-115

ISBN: 978-989-758-057-4

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

2 METHODS

2.1 Hardware

To perform the experiments, a 3D accelerometer

with a MMB Sensor v.1.0.9 (Multimediabox,

Netherlands) was placed around the right wrist of

the subjects (X axis oriented towards hand, Y on a

side and Z upright), as shown in Figure 1. The

sensor was placed in the waterproof plastic casing,

which was then attached to the wrist using watch

strap. The dimensions of sensor are 3cm x 3cm, with

a weight of 33g providing an easy-to-use measuring

system. The sensor had a sampling rate of 100Hz.

To provide a constant communication with a

computer, the data were wirelessly transferred via

USB receiver.

Figure 1: Scheme presenting 3D accelerometer with the

orientation of its axes when attached to the right hand.

2.2 Data Collection

For this study we collected the data from 10

swimmers (all European Youth Championship

swimming level) during their regular training

performance in the sport school in Antwerp,

Belgium. The baseline characteristics of the subjects

are shown in Table 1.

Table 1: Subjects characteristics (mean ± standard

deviation).

Males (n=5) Females (n=5)

Age (years) 16.8 ± 0.6 15.9 ± 0.9

Height (cm) 182 ± 4 172 ± 7

Weight (kg) 74 ± 4 64 ± 3

During the development stage, data from 3

swimmers (2 males and 1 female) were collected for

algorithm development. Subsequently, data were

additionally collected to validate the findings (data

of 7 swimmers). All the trainings were performed in

an Olympic size pool of 50m. Each of the swimmers

trained according to a predefined protocol: 200m

medley (combination of all 4 swimming styles:

crawl, breaststroke, backstroke and butterfly) of

their regular training pace which is then followed by

a 200m of same pace and 50m of fast pace for the

each swimming style separately. Laps were labelled

and measured in parallel with a hand stopwatch.

2.3 Data Processing

The algorithm development was done in an offline

framework in MATLAB (7.14, The MathWorks,

Natick, Massachusetts). Figure 2 presents data taken

from the X axis of the accelerometer at 200m

medley swimming. The raw signal (Figure 2, panel

A) from each axis was initially filtered using a low

pass filter with a cut-off frequency of 2Hz (Figure 2,

panel B). Visual differences between swimming

styles and turning were already apparent. Based on a

trial and error method, we determined thresholds for

acceleration on X and Y axis that have to be reached

to recognize swimming (shown in Figure 3). In

panel C we indicate with 0 when a period is

recognized as not swimming, while with 1 when it is

recognized as swimming. Pulse train mainly comes

from each stroke that the swimmer made, hereof

high frequency of pulses. Evidently, most of zeros

are false negative, thus if the time difference

between two pulses is less than 2.5s they are being

merged (shown in panel D, Figure 2).

To determine the style, in addition to already

identified swimming, new thresholds were defined.

Flow chart for swimming and stroke recognition is

shown in Figure 3. To detect crawl and butterfly,

average on X axis had to be lower than 0.7G.

Moreover, to distinguish between them, minimum of

acceleration on Y axis had to be lower/higher than

0.8G. In contrast, for breaststroke and backstroke,

icSPORTS2014-InternationalCongressonSportSciencesResearchandTechnologySupport

112

Figure 2: Obtained data from X axis of accelerometer during 200m medley; Panel A/ Raw signal; B/ Filtered signal with

labels; C/ Swimming (1) or not (0) upon applying thresholds on the X and Y axis; D/ Swimming sessions after correction.

OnlineMonitoringofSwimmerTrainingUsinga3DAccelerometer-IdentifyingSwimmingandSwimmingStyle

113

besides X axis threshold average on Y axis as

positive/negative was observed for style recognition.

To validate the algorithm, simulation of online

monitoring was performed. New data were released

to run through the algorithm continuously. The first

window was 20s long which was sliding over the

data.

3 RESULTS

For the 7 validation subjects, we had in total 42 laps

of swimming, each was 50m long. Our algorithm

showed highest performance by identifying all 42

swimming intervals and all 42 turning points.

Moreover, we achieved high accuracy when it

comes to detecting different swimming styles (Table

2). The algorithm detected all crawl, breaststroke

and backstroke laps, however identifying butterfly

failed in 2 out of 42 laps. Taken together, true

positive rate reaches 99%.

Table 2: Stroke recognition (TP = True positive; FP =

False positive, TN = True negative; FN = False negative).

TP FP TN FN

Crawl 42/42 1/126 125/126 0/42

Breaststroke 42/42 1/126 125/126 0/42

Backstroke 42/42 0/126 126/126 0/42

Butterfly 40/42 0/126 126/126 2/42

Figure 3: The flow chart of the stroke recognition process,

together with used cut-off’s.

4 CONCLUSIONS

In this study we are demonstrating that swimming

events together with swimming styles may be

registered with high certainty using wireless 3D

accelerometers. Moreover, we show that the whole

process could be performed in online mode, which

could be beneficial for coaches when overviewing

training performance.

More outputs, such as stroke frequency, stroke

length, stroke duration etc., are to be developed.

Identifying those variables would complete the

online monitoring tool, providing the coach all the

necessary information to analyse and improve

swimmers training and performance. In practice,

whether such a system could be useful to the

coaches is not yet explored. However, having a

constant feedback on a swimming effort and

quantification of each motion can only provide

sufficient measurements for more efficient

performance (Smith et al. 2002). As addition,

suchlike monitoring tool should not be limited to

one swimmer, but secure a multi-swimmers

monitoring feature.

Substantial improvement of the developed

algorithm could be achieved by applying device and

algorithm on larger number of subjects. This would

either provide stronger validation of the monitoring

tool, or result in fine tuning of the used cut-off’s

which should again confirm high accuracy of the

algorithm for monitoring swimming performance.

ACKNOWLEDGEMENTS

We thank Brigitte Becque and swimmers from sport

school Leonadro Lyceum in Antwerp for the help

with the data collection.

REFERENCES

Auvinet, B., Gloria, E., Renault, G., & Barrey, E. 2002.

Runner's stride analysis: comparison of kinematic and

kinetic analyses under field conditions. Science &

Sports, 17, (2) 92-94

Bachlin, M. & Troster, G. 2012. Swimming performance

and technique evaluation with wearable acceleration

sensors. Pervasive and Mobile Computing, 8, (1) 68-

Jensen, U., Prade, F., & Eskofier, B. M. Classification of

kinematic swimming data with emphasis on resource

consumption, In Body Sensor Networks, IEEE, pp. 1-

icSPORTS2014-InternationalCongressonSportSciencesResearchandTechnologySupport

114

Le Sage, T., Bindel, A., Conway, P., Justham, L.,

Slawson, S., & West, A. Development of a real time

system for monitoring of swimming performance, In

8th Conference of the International Sports

Engineering Association, 2 edn, pp. 2707-2712.

Maglischo, E. 1993. Swimming fastest Mayfield Pub. Co.

Shumei, Z., McCullagh, P., Nugent, C., & Huiru, Z.

Activity Monitoring Using a Smart Phone's

Accelerometer with Hierarchical Classification, In

Intelligent Environments, IEEE, pp. 158-163.

Siirtola, P., Laurinen, P., Roning, J., & Kinnunen, H.

Efficient accelerometer-based swimming exercise

tracking, In Computational Intelligence and Data

Mining, pp. 156-161.

Smith, D.J., Norris, S.R., & Hogg, J.M. 2002.

Performance evaluation of swimmers - Scientific

tools. Sports Medicine, 32, (9) 539-554 available

from: ISI:000177265500001

Wixted, A.J., Thiel, D.V., Hahn, A.G., Gore, C.J., Pyne,

D.B., & James, D.A. 2007. Measurement of Energy

Expenditure in Elite Athletes Using MEMS-Based

Triaxial Accelerometers. Sensors Journal, IEEE, 7, (4)

481-488

OnlineMonitoringofSwimmerTrainingUsinga3DAccelerometer-IdentifyingSwimmingandSwimmingStyle

115