iBoccia

Monitoring Elderly While Playing Boccia Gameplay

Carina Figueira

, Joana Silva

, Ant

onio Santos

, Filipe Sousa

, Vin

ıcius Silva

, Jo

ao Ramos

Filomena Soares

, Paulo Novais

and Pedro Arezes

Fraunhofer Portugal Research Center for Assistive Information and Communication Solutions,

Rua Alfredo Allen, 455/461, Porto, Portugal

Algoritmi Centre, Industrial Electronics Department, University of Minho, Guimar

aes, Portugal

Algoritmi Centre, Department of Informatics, University of Minho, Braga, Portugal

Algoritmi Centre, Production Systems Department, University of Minho, Guimar

aes, Portugal

Keywords:

Activity Monitoring, Boccia, Kinect, Wearable Devices.

Abstract:

The size of the aging population has been increasing over the last years, leading to a search for solutions that

can increase the quality of life of the elderlies. One of the main means of action is focused on their physical

activity. A non-sedentary life can help in disease prevention and disability reduction, leading to an independent

living with quality. Moreover, the practice of physical exercise can decrease fall risks and its consequences.

Furthermore, it is desirable that the solutions can be accessed by anyone, with a low inherent cost. The Boccia

game is a good way to promote physical activity to the elderly, due to its simplicity and easy adaptability to

the physical limitations of the elderly. Following this trend, this paper presents iBoccia, a novel framework

to monitor elderly while playing Boccia game, through wearable sensors, Mio Fuse band and pandlet (inertial

sensor), and a non-wearable device, Kinect camera. Several performance metrics are expected to be measured

during the gameplay. Using the pandlet we calculate wrist rotation angles and force applied during ball throw,

using the Kinect we recognize facial expressions and from the Mio Fuse band we retrieve heart rate.

1 INTRODUCTION

Physical inactivity is one of the main causes of several

health diseases, as heart diseases, beyond being cor-

related to overweight and obesity (Lee et al., 2012;

Blair et al., 1999; Warburton et al., 2006). (Lee et al.,

2012) estimated that physical inactivity is the cause of

6% of the burden of disease from coronary heart dis-

ease, 7% of type 2 diabetes, 10% of breast cancer and

10% of colon cancer. Moreover, it is the cause of 9%

of premature deaths, causing the death of more than

5.3 million of the 57 million deaths occurred in 2008

worldwide (Lee et al., 2012). The practice of physi-

cal exercises may increase cardiorespiratory and mus-

cular ﬁtness, functional health, improve bones and

joint health, and cognitive functions (Warburton et al.,

2006; Lee et al., 2012).

The physical inactivity tends to become more pro-

nounced while aging, making the elderly the most

sedentary age group. Therefore, it is important to ﬁnd

solutions that may solve this problem, encouraging

physical activity practice. For this, it is essential to

give the elderly the necessary motivation, with pleas-

ant, social and fun solutions, achieved through games

in order to promote social activity and interaction, and

improve self-conﬁdence and quality of life. Boccia is

a simple game that can be adapted to the age and limi-

tations, and beyond the practice of physical activity, it

promotes the contact with others. Therefore, this pa-

per discusses the possibility of monitoring a group of

elderly people playing the boccia game. The objective

is to collect performance and affective information as

well as movements data.

For the elderly it can be seen as a way to moni-

tor the game through the suggestion of performance

improvement of some of the movements. For formal

caregivers besides the playful aspect, it may be in-

teresting for them to realize what kind of moves were

made during the game, access the affective state of the

patient and to detect physical or cognitive declines by

analyzing the data collected.

For the system development it will be important

670

Figueira, C., Silva, J., Santos, A., Sousa, F., Silva, V., Ramos, J., Soares, F., Novais, P. and Azeres, P.

iBoccia - Monitoring Elderly While Playing Boccia Gameplay.

DOI: 10.5220/0006473606700675

In Proceedings of the 14th International Conference on Informatics in Control, Automation and Robotics (ICINCO 2017) - Volume 1, pages 670-675

ISBN: 978-989-758-263-9

to collect acceleration data of the arm of the player as

well as through the Kinect to extract facial cues, de-

termine the angles, and to analyze movements during

the game. In a ﬁrst phase the game results of each

player will be manually inserted into the system and

it may help in suggesting improvements in the move-

ment of the ball. Subsequently, the system can be im-

proved by incorporating an analysis of the play and

rating through processing and analysis the position of

the ball in the game.

This article is organized as follows: in section 2

the Boccia game scenario is explained. Section 3

presents the state of the art solutions for monitor phys-

ical activity. The system architecture is presented in

section 4. The article ﬁnishes with section 5 concern-

ing the conclusions and future work.

2 BOCCIA GAME

Boccia is an indoor precision game played with one

white, six blue and six red leather balls. The objec-

tive is to score by throwing the colored balls as close

as possible to the white one (called jack). It is allowed

to throw the balls with hands, kicked with feet or, in

a case of a severe disability, launch with aid instru-

ments. This modality can be played individually, by

pairs or by teams (Fong et al., 2012; BISFed, 2016).

Originally designed for individuals with cerebral

palsy, Boccia has become a Paralympic Game in

1984, being played by athletes with different physi-

cal and functional disabilities (Fong et al., 2012; BIS-

Fed, 2016). Apart from its competitive side, Boccia

game can also be used as an alternative to rehabilita-

tion exercises, as well as be adapted to be played by

the elderly, encouraging the physical activity practice,

assisting in balance and coordination and increasing

strength and ﬂexibility. Moreover, it can help in self-

conﬁdence and appreciation, as well as self-esteem

improvement. In addition, it helps the player to de-

velop character and to understand social norms, by

playing as a team.

3 STATE OF THE ART

With a large application areas, as health, ﬁtness and

safety, the demand for physical activity monitoring

has been increasing, following the technological ad-

vances (Choudhury et al., 2008; Figueira et al., 2015).

Herewith, several solutions have been commercial-

ized, with increasingly features and functionalities.

These solutions are mostly based on wearable or

smartphone built-in sensors, and are focused on ﬁt-

ness, calculating the number of calories lost, steps

given and distance traveled, on safety, with emer-

gency buttons to call for help, or on physiological

ﬁeld, monitoring the heart rate and body temperature,

for example (Guo et al., 2013; Mukhopadhyay, 2015).

Regarding targeted systems for seniors, several

systems have been developed in order to measure

their movements and discriminate between everyday

movements and emergencies, like being laying down

or falling. These systems are suitable for elderly mon-

itoring living at home or institutions (Charlon et al.,

2013; Van Kasteren et al., 2010; Ohta et al., 2002).

Gociety has two physical activity and fall monitor-

ing solutions, targeted primarily for the elderly: the

GoLivePhone and the GoLiveWear. GoLivePhone is

an Android application that detects falls and evalu-

ates its associated risks. When a fall is detected an

emergency message is sent to the caregiver. It also

has an emergency button, where one touch is enough

to alert professionals and caregivers. It also notiﬁes

about taking medication and collects physical activ-

ity data (number of steps taken, calories lost and time

spent in different activities such as walking or stand-

ing). In addition, caregivers can also receive notiﬁ-

cations when the user is inactive for a certain period

of time. GoLiveWear is a waterproof clip-on device

with a built-in emergency button that, connected to

the GoLivePhone application, can access to all of its

features without the need of having the smartphone al-

ways around (Gociety, 2016). Among other features,

the AT&T’s EverThere Emergency Response System

has also a call button that notiﬁes caregivers if the

senior experiences a fall or other emergency (Ever-

There, 2017) and the Flowie solution is a persuasive

tool which aims to encourage the elderly to walk more

(Albaina et al., 2009).

Despite all the existing solutions, there is none

adapted to a gaming environment, monitoring the el-

derly while playing, for example, a Boccia gameplay.

4 SYSTEM ARCHITECTURE

The architecture was designed envisioning the in-

tegration of different sensors to monitor the game

movements and the face expressions of the user in or-

der to suggest improvements of the game movements

to the players. The system will also provide and log

this sensor data and provide it to caregivers to assist

them in the detection of physical or cognitive decline.

In Figure 1 the system architecture is presented.

The movements and acceleration of the user are

collected using a sensor in the wrist called pandlet.

iBoccia - Monitoring Elderly While Playing Boccia Gameplay

671

Figure 1: iBoccia architecture.

Additionally, a kinect is used to collect the gestures,

body position, posture and face expressions of the

user. On the wrist, another sensor, the Mio Fuse, col-

lects the heart rate of the user. The pandlet and Mio

Fuse communicates by employing the Bluethooth 4.0

low energy (BLE) protocol, which is an industry-

standard and allows devices to run for long periods.

The kinect is connected through an USB cable to the

Intel Compute Stick which is used to retrieve the data

from the different sensors and display in the LCD

this information to the user. The software architec-

ture is based on a three-tier framework that organizes

modules (games) into dynamic categories, acting as a

launcher and providing them a set of APIs for com-

mon functionalities, such as sensor communication

(HRate lib, Sensor lib, motion lib) user proﬁle and

game session management. The proposed system ar-

chitecture allows the modules to be agnostic on which

sensors are currently supported and being used, as

well as which backend server may be used to man-

age user proﬁles and store game session data. The

API (Matos et al., 2014) for sensor communication

provided by the framework is implemented as shared

library/DLL in pure native code (C++) and supports

most of the existing platforms: Windows, Linux and

Android. The library is responsible for all communi-

cation details with any sensor. The received data from

sensors is processed and merged into a single proto-

col and send to any client (game) that asked for it and

provided a callback. The protocol considers data from

inertial sensors, gestures, body tracking, hand track-

ing, face’s expressions, and heart rate. The process-

ing algorithms use this for gesture recognition and fa-

cial expression recognition. The gestures recognition

algorithm is used to evaluate the performance of the

player movement and suggest improvements based on

the game classiﬁcation. The facial expression recog-

nition algorithm is used to detect emotions, opinions,

and clues regarding player’s cognitive states. The next

sections include further detail about the sensors that

are used and also the algorithms adopted to achieved

the proposed goals.

4.1 Sensors

The main purpose of this project is to be used in

community or nursing home unsupervised contexts,

employing standard and relatively inexpensive equip-

ment to monitor elderly players during a Boccia game

scenario. Thus, currently there is support for a small

set of available sensor devices, such as the Kinect sen-

sor, a Mio Fuse, and pandlet (Figure 1). To study the

behavior of each player during the game, all the sen-

sory information is going to be combined, allowing to

actively track the player movements and his/her emo-

tional state.

4.1.1 Kinect

There is a connection between affective states and ex-

ercise in real life (Sch

ondube et al., 2016). Under-

standing this association may be important for creat-

ing successful exercise promotion programs. More-

over, the strongest factor that contributes to the main-

tenance of the exercise behavior can be the positive

affective valence (Sch

ondube et al., 2016). The user

affective state might affect exercise in daily life.

The Microsoft Kinect V2 is a depth sensor that

employs the time of light technique for tracking the

body movements, extract facial features, and recog-

nizing gestures of the user. It has a built-in a 1080p

color camera, a 3D depth sensor, and a microphone

array. A Software Development Kit (SDK) was re-

leased by Microsoft, giving access to the raw sensor

data streams as well as skeletal tracking. In order to

accomplish the goals of this work the Kinect V2 is go-

ing to be used along with its SDK and the Kinteract

(Matos et al., 2014) library to extract facial features,

ICINCO 2017 - 14th International Conference on Informatics in Control, Automation and Robotics

672

track the player, and recognizing gestures, allowing to

monitor the user activity during a Boccia game sce-

nario. The APIs for the Kinect allow to retrieve the

position (in high detail) of up to 36 facial landmarks

and it gives access to a skeleton model composed of

25 joints that can be used to track each joint position

(Microsoft, 2017). Additionally, the Kinteract library

can compute the angles enabling, together with the

Kinect SDK, the employment of gesture recognition.

4.1.2 Inertial Sensors

The Pandlet (Letting Everything Sense) is a bracelet

with a novel architecture of embedded electronics for

wireless devices that can be used to develop Wearable

and IoT solutions. It includes an accelerometer, mag-

netometer, and gyroscope that can be used to track

the user movements during the game, with a 100 Hz

frequency. Moreover, the Pandlet is composed by an

ARM M0+, running at 16 MHz and an environmen-

tal measurement unit. It is also charged wirelessly

and Bluetooth Smart. The radio range is 40 meters

in line of sight, and in a high 2.4 GHz polluted en-

vironment. Although the kinect can detect the player

movements the force applied when launching the ball

is hard to assess. Using the pandlet and by knowing

the ball mass we can easily estimate the force applied

to the ball and estimate the landing point. The pan-

dlet’s acceleration fused with the kinect’s tracking of

player’s movements will provide the required data to

the gesture recognition algorithm to access the player

performance in the game.

4.1.3 Heart Rate Monitor

Physical exercise has proven its beneﬁts in the health

of a patient. Thus, the risk of heart diseases is re-

duced. However, there are limits to the exercise which

considers the maximum heart rate (HR

max

), calcu-

lated as 220 − age in years (Atwal et al., 2002). Sur-

passing this limit may lead to severe problems due to

excessive heart stress. The heart rate is retrieved by

the Mio Fuse, which is placed on the patient’s wrist.

Through this device the caregiver may control the user

heart health and, if necessary, stop the game in an ear-

lier stage.

4.2 Processing Algorithms

4.2.1 Gestures Recognition Algorithms

Gestures recognition can be considered as a super-

vised classiﬁcation problem, if annotated gestures se-

quential data is given to a classiﬁer as a time series.

Some of the common used approaches rely on: Sup-

port Vector Machines, K-Nearest Neighbors or Na

ıve

Bayes. These algorithms use labeled data (in this

case, the gestures) to train the classiﬁer, which be-

comes able to map new inputs (Wilde, 2010). On

the other hand, for supervised learning and for an

in-depth study of gestures recognition, it has been

also considered the Hidden Markov Models (HMM),

which aims to present results from the present out-

comes through probabilities. HMM can model con-

ditions of the occasions that may happen repeatedly

over time or predictable events that take place over

time (Fosler-Lussier, 1998). Dynamic Time Warp-

ing (DTW) (Muhammad and Devi, 2016) is another

algorithm commonly used for time series similarity

evaluation, which aligns each sequence prior to es-

tablishing the distance measurement. Moreover, Re-

current Neural Networks have also been used to ges-

ture recognition (Eleni, 2015; Murakami and Taguchi,

1991).

The authors have been working in gesture recog-

nition. With this, several projects and master thesis

have concerned the classiﬁcation of gestures using

HMM, DTW or other classiﬁcation algorithms. Some

examples are the (da Silva, 2013), where hand ges-

tures where recognized with a smartphone, (da Costa

et al., 2015) classiﬁed activities of daily-living in

post-stroke patients and (Freixo, 2015) combined

electromyography and inertial sensor for gesture de-

tection and control.

The proposed system will rely on supervised algo-

rithms, that will be trained with annotated data from

the Boccia game movements. Boccia is characterized

as a throwing sport, so the movements to recognize

will focus on ball throw. Since the objective is to get

close to the jack ball, force is not always necessary to

reach the better position. So, precision is occasionally

worth than long throws. With practice, players de-

velop their own strategies to win the game, as the way

they throw the ball or roll depends on players’ body

movements. The goal of this system, is to provide

user movements’ feedback during the game, as for ex-

ample, when a user is given with speciﬁc information

of ball throw and given the ball position on the court,

users can understand which movements result better

and try to improve their performance. As the physics

mechanisms applied to the ball on the court could

not be inferred by the system, due to uncontrolled

parameters (opponent’s balls positions, jack position,

rolling on the ﬂoor, etc.), we can not relate move-

ments performance with game scoring. However, the

player is able to infer such relation and relate higher

scores with different body movements. Focusing on

game related movements, the system could be able

iBoccia - Monitoring Elderly While Playing Boccia Gameplay

673

to recognize and characterize ball throws into their

main movements: wrist movements for ball prepara-

tion, body position adjustment during court observa-

tion, wrist rotation angle and force applied during ball

throw. For the caregiver point of view, these move-

ments’ characterization could be useful to evaluate

arm movements of the user, that are important for up-

per body rehabilitation.

4.2.2 Facial Expression Recognition Algorithms

Facial expressions are innate in any communication

and interaction between humans. They can trans-

mit emotions, opinions, and clues regarding cognitive

states. Several psychological studies have been con-

ducted in order to decode the information contained

in a facial expression. For example, the system de-

veloped by Ekman and Friesen (Ekman and Friesen,

1978), the Facial Action Codding System (FACS), al-

lowed researches to analyze and classify facial ex-

pressions in a standardized framework. This system

associates the action of the muscles to the changes in

facial appearance. The measurements of the FACS

are called Action Units which are actions performed

by a muscle or a group of muscles. There are two

main approaches regarding facial expression recogni-

tion, the feature-based ones, which uses textural or

geometrical information, and the template based ones

that uses 2D or 3D head and face models (Kotsia and

Pitas, 2007). Usually, these systems try to classify

the six basic emotions, also considered the six uni-

versal emotions – happiness, sadness, anger, surprise,

fear, and disgust. Some researchers are using machine

learning techniques to detect such patterns.

(Silva et al., 2014) proposed an automatic human

facial expression recognition frame-based system that

classiﬁes six basic facial expressions plus the neu-

tral state. The proposed framework compared the

performance of three different classiﬁers (Artiﬁcial

Neural Network, Linear Discriminant Analysis and k-

Nearest Neighbor). Other authors, (Silva et al., 2016;

Youssef et al., 2013), used Support Vector Machines

(SVM) in order to classify the emotional state of the

user.

5 FINAL REMARKS

The present paper concerns the development of phys-

ical and cognitive monitoring technologies to support

and promote exercise programs, focusing in elderly

people. Nowadays, physical inactivity can be one of

the main causes of several illness. Physical inactivity

tends to be more present in the elderly. Following this

trend, the main purpose of the present work is to de-

velop a system for collecting performance and affec-

tive information as well as movements data. As stated

before, affective states should be considered in creat-

ing effective interventions to foster exercise behavior

and enhance maintenance. The system is composed

of a Kinect sensor, a Mio Fuse band, the pandlet, and

an Intel compute stick.

The developed system is going to be evaluated in

different scenarios in order to assess its performance.

This evaluation will be based on several metrics, tak-

ing into account the wrist rotation angles and force

applied during the ball throw calculated by the Pan-

dlet. Through the Kinect, facial expressions recogni-

tion will also be performed and with the Mio Fuse

band the heart rate will be retrieve, too. First, the

system is going to be tested in a laboratorial envi-

ronment with a set of 12 participants, simulating a

Boccia game. This ﬁrst evaluation step is going to be

conducted with the purpose of detecting the system

constraints and to tune the conditions of the experi-

mental scheme. Several Boccia movements should be

deﬁned and validated by the proposed system. Also,

emotional states should be simulated by the test group

and correctly detected and correlated by the system.

After this ﬁrst validation in a controlled environment,

a second test should be developed in a real-world con-

text. So, an evaluation will be performed with a set of

12 elderly participants in a nursing room. In this eval-

uation, the participants are going to be monitored on

site by professionals, and the research team is going

to oversee the progress of the game, and monitor the

system. The detection of the movements and the de-

gree of engagement in the game should be correctly

monitored. iBoccia will then be subjected to exten-

sive tests in order to validate it as an adequate tool to

promote physical activity in the elderly.

ACKNOWLEDGEMENTS

We would like to acknowledge the ﬁnancial support

obtained from North Portugal Regional Operational

Programme (NORTE 2020), Portugal 2020 and the

European Regional Development Fund (ERDF) from

European Union through the project Symbiotic tech-

nology for societal efﬁciency gains: Deus ex Machina

(DEM), NORTE-01-0145-FEDER-000026.

REFERENCES

Albaina, I. M., Visser, T., van der Mast, C. A., and Vas-

tenburg, M. H. (2009). Flowie: A persuasive virtual

ICINCO 2017 - 14th International Conference on Informatics in Control, Automation and Robotics

674

coach to motivate elderly individuals to walk. In Per-

vasive Computing Technologies for Healthcare, 2009.

PervasiveHealth 2009. 3rd International Conference

on, pages 1–7. IEEE.

Atwal, S., Porter, J., and MacDonald, P. (2002). Cardiovas-

cular effects of strenuous exercise in adult recreational

hockey: The hockey heart study. CMAJ.

BISFed (2016). Bisfed - boccia international sports federa-

tion. http://www.bisfed.com/, [Online; accessed 2017-

04-10].

Blair, S. N., Brodney, S., et al. (1999). Effects of physi-

cal inactivity and obesity on morbidity and mortality:

current evidence and research issues. Medicine and

science in sports and exercise, 31:S646–S662.

Charlon, Y., Bourennane, W., Bettahar, F., and Campo, E.

(2013). Activity monitoring system for elderly in a

context of smart home. IRBM, 34(1):60–63.

Choudhury, T., Consolvo, S., Harrison, B., Hightower, J.,

LaMarca, A., LeGrand, L., Rahimi, A., Rea, A., Bor-

dello, G., Hemingway, B., et al. (2008). The mobile

sensing platform: An embedded activity recognition

system. IEEE Pervasive Computing, 7(2).

da Costa, M. J. M. P. et al. (2015). Mobile real-time classi-

ﬁcation of activities of daily-living in post-stroke pa-

tients.

da Silva, P. R. D. C. (2013). Smartphone gesture learning.

Ekman, P. and Friesen, W. V. (1978). Facial Action Coding

System: A Technique for the Measurement of Facial

Movement.

Eleni, T. (2015). Gesture Recognition with a Convolutional

Long Short Term Memory Recurrent Neural Network.

PhD thesis.

EverThere, A. (2017). How everthere works. https://www.

att.com/att/InnovationStore/images/products/everthere

/docs/HowItWorks.pdf, [Online; accessed 2017-04-

07].

Figueira, C., Matias, R., and Gamboa, H. (2015). Body

location independent activity monitoring. In BIOSIG-

NALS, page 8.

Fong, D. T.-P., Yam, K.-Y., Chu, V. W.-S., Cheung, R. T.-H.,

and Chan, K.-M. (2012). Upper limb muscle fatigue

during prolonged boccia games with underarm throw-

ing technique. Sports Biomechanics, 11(4):441–451.

Fosler-Lussier, E. (1998). Markov models and hidden

markov models: a brief tutorial. International Com-

puter Science Institute.

Freixo, R. A. E. (2015). Electromyography and inertial

sensor-based gesture detection and control.

Gociety (2016). Gociety solutions. http://www.gociety so-

lutions.com/, [Online; accessed 2017-04-11].

Guo, F., Li, Y., Kankanhalli, M. S., and Brown, M. S.

(2013). An evaluation of wearable activity monitoring

devices. In Proceedings of the 1st ACM international

workshop on Personal data meets distributed multi-

media, pages 31–34. ACM.

Kotsia, I. and Pitas, I. (2007). Facial expression recognition

in image sequences using geometric deformation fea-

tures and support vector machines. IEEE Transactions

on Image Processing, 16(1):172–187.

Lee, I.-M., Shiroma, E. J., Lobelo, F., Puska, P., Blair,

S. N., Katzmarzyk, P. T., Group, L. P. A. S. W.,

et al. (2012). Effect of physical inactivity on major

non-communicable diseases worldwide: an analysis

of burden of disease and life expectancy. The lancet,

380(9838):219–229.

Matos, N., Santos, A., and Vasconcelos, A. (2014). Kin-

teract: A multi-sensor physical rehabilitation solution

based on interactive games. In Communications in

Computer and Information Science.

Microsoft (2017). HighDetailFacePoints Enumera-

tion. https://msdn.microsoft.com/en-us/library/

microsoft.kinect.face.highdetailfacepoints.aspx,

[Online; accessed 2017-04-10].

Muhammad, P. and Devi, S. A. (2016). Hand gesture user

interface for smart devices based on mems sensors.

Procedia Computer Science, 93:940 – 946.

Mukhopadhyay, S. C. (2015). Wearable sensors for human

activity monitoring: A review. IEEE sensors journal,

15(3):1321–1330.

Murakami, K. and Taguchi, H. (1991). Gesture recognition

using recurrent neural networks. In Proceedings of the

SIGCHI conference on Human factors in computing

systems, pages 237–242. ACM.

Ohta, S., Nakamoto, H., Shinagawa, Y., and Tanikawa, T.

(2002). A health monitoring system for elderly peo-

ple living alone. Journal of telemedicine and telecare,

8(3):151–156.

Sch

ondube, A., Kanning, M., and Fuchs, R. (2016).

The bidirectional effect between momentary affective

states and exercise duration on a day level. Frontiers

in Psychology, 7:1414.

Silva, C., Sobral, A., and Vieira, R. T. (2014). An auto-

matic facial expression recognition system evaluated

with different classiﬁers. X Workshop de Vis

ao Com-

putacional (WVC’2014), pages 208–212.

Silva, V., Soares, F., Esteves, J. S., Figueiredo, J., Le

ao,

C. P., Santos, C., and Paula, A. (2016). Real-

time Emotions Recognition System. In 8th Inter-

national Congress on Ultra Modern Telecommunica-

tions and Control Systems and Workshops (ICUMT),

pages 201–206, Lisboa.

Van Kasteren, T., Englebienne, G., and Kr

ose, B. J. (2010).

An activity monitoring system for elderly care using

generative and discriminative models. Personal and

ubiquitous computing, 14(6):489–498.

Warburton, D. E., Nicol, C. W., and Bredin, S. S. (2006).

Health beneﬁts of physical activity: the evidence.

Canadian medical association journal, 174(6):801–

809.

Wilde, A. G. (2010). An overview of human activity de-

tection technologies for pervasive systems. Depart-

ment of Informatics University of Fribourg, Switzer-

land, 212.

Youssef, A. E., Aly, S. F., Ibrahim, A. S., and Abbott,

a. L. (2013). Auto-Optimized Multimodal Expres-

sion Recognition Framework Using 3D Kinect Data

for ASD Therapeutic Aid. International Journal of

Modeling and Optimization, 3(2):112–115.

iBoccia - Monitoring Elderly While Playing Boccia Gameplay

675