Motion Recognition

This research focuses on sensing context, modeling human behavior and developing a new architecture for a cognitive phone platform. We combine the latest positioning technologies and phone sensors to capture human movements in natural environments and use the movements to study human behavior. Contexts in this research are abstracted as a Context Pyramid which includes six levels: Raw Sensor Data, Physical Parameter, Features/Patterns, Simple Contextual Descriptors, Activity-Level Descriptors, and Rich Context. To achieve implementation of the Context Pyramid on a cognitive phone, three key technologies are utilized: ubiquitous positioning, motion recognition, and human behavior modeling. Preliminary tests indicate that we have successfully achieved the Activity-Level Descriptors level with our LoMoCo (Location-Motion-Context) model. Location accuracy of the proposed solution is up to 1.9 meters in corridor environments and 3.5 meters in open spaces. Test results also indicate that the motion states are recognized with an accuracy rate up to 92.9% using a Least Square-Support Vector Machine (LS-SVM) classifier.

The paper presents an indoor navigation solution by combining physical motion recognition with wireless positioning. Twenty-seven simple features are extracted from the built-in accelerometers and magnetometers in a smartphone. Eight common motion states used during indoor navigation are detected by a Least Square-Support Vector Machines (LS-SVM) classification algorithm, e.g., static, standing with hand swinging, normal walking while holding the phone in hand, normal walking with hand swinging, fast walking, U-turning, going up stairs, and going down stairs. The results indicate that the motion states are recognized with an accuracy of up to 95.53% for the test cases employed in this study. A motion recognition assisted wireless positioning approach is applied to determine the position of a mobile user. Field tests show a 1.22 m mean error in -Static Tests‖ and a 3.53 m in -Stop-Go Tests‖.

This research focuses on sensing context, modeling human behavior and developing a new architecture for a cognitive phone platform. We combine the latest positioning technologies and phone sensors to capture human movements in natural environments and use the movements to study human behavior. Contexts in this research are abstracted as a Context Pyramid which includes six levels: Raw Sensor Data, Physical Parameter, Features/Patterns, Simple Contextual Descriptors, Activity-Level Descriptors, and Rich Context. To achieve implementation of the Context Pyramid on a cognitive phone, three key technologies are utilized: ubiquitous positioning, motion recognition, and human behavior modeling. Preliminary tests indicate that we have successfully achieved the Activity-Level Descriptors level with our LoMoCo (Location-Motion-Context) model. Location accuracy of the proposed solution is up to 1.9 meters in corridor environments and 3.5 meters in open spaces. Test results also indicate that the motion states are recognized with an accuracy rate up to 92.9% using a Least Square-Support Vector Machine (LS-SVM) classifier.

The global population of older adults has increased, leading to a rising number of older adults in nursing homes without
adequate care. This study proposes a smart wearable device for detecting and classifying abnormal behaviour in older adults
in nursing homes. The device utilizes artificial intelligence technology to detect abnormal movements through behavioural
data collection and target positioning. The intelligent recognition system and hardware sensors were tested using cloud
computing and wireless sensor networks (WSNs), comparing their performance with other technologies through
simulations. A triple-axis acceleration sensor collected motion behaviour data, and Zigbee enabled the wireless transfer of
the sensor data. The Backpropagation (BP) neural network detected and classified abnormal behaviour based on simulated
sensor data. The proposed smart wearable device offers indoor positioning, detection, and classification of abnormal
behaviour. The embedded intelligent system detects routine motions like walking and abnormal behaviours such as falls. In
emergencies, the system alerts healthcare workers for immediate safety measures. This study lays the groundwork for
future AI-based technology implementation in nursing homes, advancing care for older adults.

The growing demand for physical rehabilitation processes can result in the rising of costs and waiting lists, becoming a threat to healthcare services’ sustainability. Telerehabilitation solutions can help in this issue by discharging patients from points of care while improving their adherence to treatment. Sensing devices are used to collect data so that the physiotherapists can monitor and evaluate the patients’ activity in the scheduled sessions. This paper presents a software platform that aims to meet the needs of the rehabilitation experts and the patients along a physical rehabilitation plan, allowing its use in outpatient scenarios. It is meant to be low-cost and easy-to-use, improving patients and experts experience. We show the satisfactory results already obtained from its use, in terms of the accuracy evaluating the exercises, and the degree of users’ acceptance. We conclude that this platform is suitable and technically feasible to carry out rehabilitation plans outsid...

Effective classification and detection of equipment on construction sites is critical for efficient equipment management. Despite substantial research efforts in this field, most previous studies have focused on classifying a limited number of equipment categories. Furthermore, there is a scarcity of research dedicated to heavy construction equipment. Hence, this study develops a robust Convolutional Neural Network (CNN) model to classify heavy construction machinery into 12 different types. The study utilizes a comprehensive dataset of equipment images, which was divided into three distinct subsets: 60% for training the model, 30% for validating its performance, and 10% for testing its accuracy. The model’s robustness was ensured by monitoring accuracy and loss measures during the training and validation phases. The CNN model achieved approximately 85% training accuracy with a minimum loss of 0.40. The testing phase revealed a high overall precision of 80%. The CNN model accurately classifies concrete mixer machines and telescopic handlers with an Area Under the Curve (AUC) of 0.92, however pile driving machines have a lower accuracy with an AUC of 0.83. These findings demonstrate the model’s high ability to distinguish between several types of heavy construction equipment. This paper contributes to the relatively unexplored area of classifying heavy construction equipment by providing a practical tool for automating equipment classification, leading to enhanced efficiency, safety, and maintenance protocols in construction management.

The paper presents an indoor navigation solution by combining physical motion recognition with wireless positioning. Twenty-seven simple features are extracted from the built-in accelerometers and magnetometers in a smartphone. Eight common motion states used during indoor navigation are detected by a Least Square-Support Vector Machines (LS-SVM) classification algorithm, e.g., static, standing with hand swinging, normal walking while holding the phone in hand, normal walking with hand swinging, fast walking, U-turning, going up stairs, and going down stairs. The results indicate that the motion states are recognized with an accuracy of up to 95.53% for the test cases employed in this study. A motion recognition assisted wireless positioning approach is applied to determine the position of a mobile user. Field tests show a 1.22 m mean error in-Static Tests‖ and a 3.53 m in-Stop-Go Tests‖.

Where a licence is displayed above, please note the terms and conditions of the licence govern your use of this document. When citing, please reference the published version. Take down policy While the University of Birmingham exercises care and attention in making items available there are rare occasions when an item has been uploaded in error or has been deemed to be commercially or otherwise sensitive.

Construction workers are commonly subjected to ergonomic risks due to manual material handling that requires high levels of energy input over long work hours. Fatigue in musculature is associated with decline in postural stability, motor performance, and altered normal motion patterns, leading to heightened risks of work-related musculoskeletal disorders. Physical fatigue has been previously demonstrated to be a good indicator of injury risks, thus, monitoring and detecting muscle fatigue during strenuous work may be advantageous in mitigating these risks. Currently, few researchers have investigated how physical fatigue and exertion can be continuously monitored for practical use outside laboratory settings. Exercise-induced fatigue has been shown to impact motor control; thus, it can be measured using jerk, the time derivative of acceleration. This paper investigates the application of a machine learning approach, Support Vector Machine (SVM), to automatically recognize jerk changes due to physical exertion. We hypothesized that physical exertion and fatigue will influence motions and thus, can be classified based on jerk values. The motion data of six expert masons were collected using IMU sensors during two bricklaying tasks. The pelvis, upper arms, and thighs jerk values were used to classify inter-and intra-subject rested and exerted states. Our results show that on average, intra-subject classification achieved an accuracy of 94% for a five-course wall building experiment and 80% for a first-course experiment, leading us to conclude that jerk changes due to physical exertion can be detected using wearable sensors and SVMs.

Manual load carrying without sufficient rest may cause work-related musculoskeletal disorders (WMSDs) and needs to be monitored at construction sites. While previous studies have been able to predict load-carrying modes using multiple wearable inertial measurement unit (IMU) sensors, wearing multiple sensors obtrudes on workers during various construction tasks. In this context, by using a single IMU sensor, this research proposes an automatic detecting technique for excessive carrying-load (DeTECLoad) to predict load-carrying weights and postures simultaneously. DeTECLoad converts the IMU data into image data using a Gramian Angular Field, and then uses a hybrid Convolutional Neural Network-Long Short-Term Memory to classify load-carrying modes from the image data. DeTECLoad provides 92.46% and 96.33% accuracies for the load-carrying weight and posture classifications, respectively. By exploiting DeTECLoad, a construction worker's excessive load-carrying tasks could be managed in situ, helping to prevent construction site WMSDs.

I want to thank Professor Kari Halonen for the opportunity to work with interesting topics in his research group and providing me with an intriguing topic for my master's thesis. I also want to thank my two advisors, post doctoral researcher Elina Ilén for her insight in textiles and assistant professor Ivan Vujaklija for his expertise in electromyography and data classification. A big thank you to my colleague Samu Järvinen as well for collaborating with me in this project. I also had a great deal of help and support from my spouse Lauri who has an abundance of medical literature and knowledge and my sister Elsa who lent me a hand when mine was not working.

Human Activity Recognition (HAR) plays an important role in behavior analysis, video surveillance, gestures recognition, gait analysis, and posture recognition. Given the recent progress of Artificial Intelligence (AI) applied to HAR, the inputs that are the data from wearable sensors can be treated as time-series from which movement events can be classified with high accuracy. In this study, a dataset of raw sensor data served as input to four different deep learning networks (DNN, CNN, LSTM, and CNN-LSTM). Differences in accuracy and learning time were then compared and evaluated for each model. An analysis of HAR was made based on an attempt to classify three activities: walking, sit-to-stand, and squatting. We also compared the performance of two different sensor data types: 3-axis linear acceleration measured from two inertial measurement units (IMUs) versus 3D acceleration of two retro-reflective markers from the high-end optoelectronic motion capture system (MOCAP). The datas...

The research context of this article is the recognition and description of dynamic textures. In image processing, the wavelet transform has been successfully used for characterizing static textures. To our best knowledge, only two works are using spatio-temporal multiscale decomposition based on tensor product for dynamic texture recognition. One contribution of this article is to analyse and compare the ability of the 2D+T curvelet transform, a geometric multiscale decomposition, for characterizing dynamic textures in image sequences. Two approaches using the 2D+T curvelet transform are presented and compared using three new large databases. A second contribution is the construction of these three publicly available benchmarks of increasing complexity. Existing benchmarks are either too small, not available or not always constructed using a reference database. Feature vectors used for recognition are described

This research focuses on sensing context, modeling human behavior and developing a new architecture for a cognitive phone platform. We combine the latest positioning technologies and phone sensors to capture human movements in natural environments and use the movements to study human behavior. Contexts in this research are abstracted as a Context Pyramid which includes six levels: Raw Sensor Data, Physical Parameter, Features/Patterns, Simple Contextual Descriptors, Activity-Level Descriptors, and Rich Context. To achieve implementation of the Context Pyramid on a cognitive phone, three key technologies are utilized: ubiquitous positioning, motion recognition, and human behavior modeling. Preliminary tests indicate that we have successfully achieved the Activity-Level Descriptors level with our LoMoCo (Location-Motion-Context) model. Location accuracy of the proposed solution is up to 1.9 meters in corridor environments and 3.5 meters in open spaces. Test results also indicate that the motion states are recognized with an accuracy rate up to 92.9% using a Least Square-Support Vector Machine (LS-SVM) classifier.

Construction workers are commonly subjected to ergonomic risks due to manual material handling that requires high levels of energy input over long work hours. Fatigue in musculature is associated with decline in postural stability, motor performance, and altered normal motion patterns, leading to heightened risks of work-related musculoskeletal disorders. Physical fatigue has been previously demonstrated to be a good indicator of injury risks, thus, monitoring and detecting muscle fatigue during strenuous work may be advantageous in mitigating these risks. Currently, few researchers have investigated how physical fatigue and exertion can be continuously monitored for practical use outside laboratory settings. Exercise-induced fatigue has been shown to impact motor control; thus, it can be measured using jerk, the time derivative of acceleration. This paper investigates the application of a machine learning approach, Support Vector Machine (SVM), to automatically recognize jerk changes due to physical exertion. We hypothesized that physical exertion and fatigue will influence motions and thus, can be classified based on jerk values. The motion data of six expert masons were collected using IMU sensors during two bricklaying tasks. The pelvis, upper arms, and thighs jerk values were used to classify inter-and intra-subject rested and exerted states. Our results show that on average, intra-subject classification achieved an accuracy of 94% for a five-course wall building experiment and 80% for a first-course experiment, leading us to conclude that jerk changes due to physical exertion can be detected using wearable sensors and SVMs.

Although heavy equipment is an indispensable resource in many construction projects, it is often underutilized. Inefficient usage patterns and frequent idling contribute to increased emissions and project costs. Efforts to improve usage patterns often begin with activity tracking. Recent research into automated activity tracking has leveraged sensing devices and Internet-of-Things (IoT) frameworks to power machine learning models that can predict the behaviors of monitored equipment. However, shallow machine learning models require complex manual feature engineering that could be further automated with more recent deep learning approaches. Deep learning approaches not only increase automation but also promise improved accuracies by avoiding biases introduced by manual feature design. This paper proposes a construction equipment activity recognition framework that uses deep learning architectures to predict the activities of heavy construction equipment monitored via accelerometers and applies this framework to a roller compactor and an excavator performing real work. The performance of a simple baseline convolutional neural network (CNN) is compared to a hybrid network that contains both convolutional and recurrent long short-term memory (LSTM) layers. The hybrid model outperforms the baseline model in all instances studied. In the task of classifying the activities of the roller compactor, the hybrid model achieves a validation accuracy of 77.1% when presented with six activities and a validation accuracy of 96.2% when distinguishing only direction. In the task of classifying seven activities of the excavator, the hybrid model achieves a validation accuracy of 77.6%, with some confusion between isolated activities and a Various category that includes elements of the isolated activities. With the Various category removed, the hybrid model achieves a validation accuracy of 90.7%. This study demonstrates that deep learning frameworks can model the activities of construction equipment with high accuracy. In particular, this work shows that convolutional and LSTM layers can each form effective parts of deep learning models that characterize equipment activities based on accelerometer data, and furthermore that these components can produce more effective models when combined. The findings of this study can be leveraged by researchers and industry professionals to develop reliable automated activity recognition systems for tracking and monitoring equipment performance and for measuring the productivity and the efficiency of the work performed.

Automated motion detection and tracking is a challenging task in traffic surveillance. In this paper, a system is developed to gather useful information from stationary cameras for detecting moving objects in digital videos. The moving detection and tracking system is developed based on optical flow estimation together with application and combination of various relevant computer vision and image processing techniques to enhance the process. To remove noises, median filter is used and the unwanted objects are removed by applying thresholding algorithms in morphological operations. Also the object type restrictions are set using blob analysis. The results show that the proposed system successfully detects and tracks moving objects in urban videos.

Lumbar-pelvic movements (LPMs) are generally performed in the clinical setting to identify limitations in a range of movements. Continuous monitoring of these movements can provide real-time feedback to both patients and medical experts with the potential of identifying activities that may precipitate symptoms of low back pain (LBP) as well as improving therapy by providing a personalised approach. Recent advances in mobile computing technology and wearable sensors have paved the way for developing mobile physical activity monitoring applications with more advanced and complex algorithms, such as deep neural network (DNN) based models. However, there is a lack of prior studies that focus on real-time LPMs detection with multimodal sensory data. Meanwhile, most research in the area of body movement detection do not consider the potential transition logic of the constituent low-level body movements (e.g., LPMs) within their corresponding high-level physical activity. This information could significantly increase accuracy of detection results. Further, current studies mainly perform deep learning-based movement detection in the cloud (or a backend server) that could increase network bandwidth and response time. To address these limitations, this paper proposes a novel LPMs detection approach using an enhanced and adapted hybrid DNN model, which includes a convolutional neural network followed by a long-short term memory recurrent network (CNN-LSTM), and performs detection locally on the mobile device. The results of a comparative evaluation of the proposed model and baseline models are described. We also introduce a set of domain-specific pre-defined rules, based on the transition logic information, to reconstruct the detection outputs to further improve the detection accuracy. INDEX TERMS Deep neural networks, domain adaptation, lumbar-pelvic movement detection, mobile platform, on-device processing, temporal-wise attention.

The technological advances in medical sensors, low-power microelectronics and miniaturization, wireless communications and networks have enabled the appearance of a new generation of wireless sensor networks: the so-called wireless body area networks (WBAN). These networks can be used for continuous monitoring of vital parameters, movement, and the surrounding environment. The data gathered by these networks contributes to improve users' quality of life and allows the creation of a knowledge database by using learning techniques, useful to infer abnormal behaviour. In this paper we present a wireless body area network architecture to recognize human movement, identify human postures and detect harmful activities in order to prevent risk situations. The WBAN was created using tiny, cheap and low-power nodes with inertial and physiological sensors, strategically placed on the human body. Doing so, in an as ubiquitous as possible way, ensures that its impact on the users' daily actions is minimum. The information collected by these sensors is transmitted to a central server capable of analysing and processing their data. The proposed system creates movement profiles based on the data sent by the WBAN's nodes, and is able to detect in real time any abnormal movement and allows for a monitored rehabilitation of the user.

The technological advances in medical sensors, low-power microelectronics and miniaturization, wireless communications and networks have enabled the appearance of a new generation of wireless sensor networks: the so-called wireless body area networks (WBAN). These networks can be used for continuous monitoring of vital parameters, movement, and the surrounding environment. The data gathered by these networks contributes to improve users' quality of life and allows the creation of a knowledge database by using learning techniques, useful to infer abnormal behaviour. In this paper we present a wireless body area network architecture to recognize human movement, identify human postures and detect harmful activities in order to prevent risk situations. The WBAN was created using tiny, cheap and low-power nodes with inertial and physiological sensors, strategically placed on the human body. Doing so, in an as ubiquitous as possible way, ensures that its impact on the users' daily actions is minimum. The information collected by these sensors is transmitted to a central server capable of analysing and processing their data. The proposed system creates movement profiles based on the data sent by the WBAN's nodes, and is able to detect in real time any abnormal movement and allows for a monitored rehabilitation of the user.

Fall detection is a major problem in the healthcare department. Elderly people are more prone to fall than others. There are more than 50% of injury-related hospitalizations in people aged over 65. Commercial fall detection devices are expensive and charge a monthly fee for their services. A more affordable and adaptable system is necessary for retirement homes and clinics to build a smart city powered by IoT and artificial intelligence. An effective fall detection system would detect a fall and send an alarm to the appropriate authorities. We propose a framework that uses edge computing where instead of sending data to the cloud, wearable devices send data to a nearby edge device like a laptop or mobile device for real-time analysis. We use cheap wearable sensor devices from MbientLab, an open source streaming engine called Apache Flink for streaming data analytics, and a long short-term memory (LSTM) network model for fall classification. The model is trained using a published dat...

Recent advancements in edge computing devices motivate us to develop a sustainable and reliable technique for multiple gait activities recognition using wearable sensors. This research work presents the multitask human walking activities recognition using human gait patterns. Human locomotion is defined as the change in the joint angles of hip, knee and ankle. To achieve the aforementioned objective, the data are collected for 50 subjects in a controlled laboratory environment using inertial measurement unit (IMU) sensors for 7 different activities. The IMU sensor is placed on the chest, left thigh, and right thigh. Total 100 samples are collected for all 7 activities. The sampling rate considered was 50 Hz. Following 7 walking activities are performed for all the 50 subjects: (i) natural walk, (ii) standing, (iii) climbing stairs, (iv) cycling, (v) jogging, (vi)running, (vii) knees bending(Crouching). The major contribution of this research paper is the design of four hybrid deep learning models to provide the generic activity recognition framework and tune the performance. The following combination of the deep learning model is designed for the classification of gait activities, namely, convolution neural network-long short-term memory (CNN-LSTM), CNN-gated recurrent unit (CNN-GRU), LSTM-CNN and LSTM-GRU. To support edge computing, the ensemble learning is utilized to optimized the model size. The proposed ensemble learning-based hybrid deep learning framework has provided a promising classification accuracy of 99.34% over other models. The other models namely CNN, LSTM, GRU, CNN-LSTM, LSTM-CNN, CNN-GRU, GRU-CNN have achieved 97.26%, 90.67%, 77.38%, 97.83%, 94.35%, 97.64%, 96.98% accuracy, respectively, on our HAG data set. The proposed technique is also validated on MHEALTH data set for comparative analysis. The hybrid deep learning model in combination with ensemble learning has outperformed other techniques. The optimized code can be used on small computation devices for walking activity recognition.

Monitoring and understanding construction workers' behavior and working conditions are essential to achieve success in construction projects. The dynamic nature of construction sites has heightened the awareness of the need for improved monitoring of the individual workers on the sites. Although several studies have shown promising results in automated motion and activity recognition using wearable motion sensors, their technical and practical feasibility was not properly validated in actual jobsites. Motion recognition models have to be evaluated in actual conditions because the motion sensor data collected in controlled conditions, and actual conditions can have different characteristics. This study proposes Long Short-Term Memory (LSTM) networks for recognizing construction workers' motions. The LSTM networks were validated through case studies in one bridge construction site and two road pavement sites. The LSTM networks showed classification accuracies of 97.6%, 95.93%, and 97.36% from three different field test sites, respectively. Through the case studies, the technical and practical feasibility of the LSTM networks was properly investigated. With the LSTM networks, it is expected that the individual workers' behavior and working conditions can be automatically monitored and managed without excessive manual observation.

This paper investigates to which degrees of playing games by using hand motions improve brain memory. Based on previous studies in psychology reporting that hand exercise could affect brain function and memory, we develop an interface for playing games by using hand motions. This interface works by detecting hand motions through image processing, translating them into commands for keyboard press, and sending such commands to the target game application; it works with any existing games without the need to modify the game source code and only requires an off-the-self webcam. An experiment is conducted on a jumping game, where three types of game interfaces-a keyboard, a Kinect (for playing with hand raising motions), and the proposed interface (for playing with hand clenching and spreading motions)-are compared. The results show that playing games using hand clenching/spreading leads to the best memory test performance according to the N-back test, a commonly used cognitive task.

This paper presents a motion-based rhythm game that facilitates rehabilitation at home. In this game, the player has to collect procedurally generated nodes, based on a pool of pre-recorded moves which can be modified by medical experts. A targeted percentage of movement can be specified for each part of the body; the game will generate a level that makes the player's movement match the targeted percentages.

We present the design outline of a context-aware interactive system for smart learning in the STEM curriculum (science, technology, engineering, and mathematics). It is based on a gameful design approach and enables “playful coached learning” (PCL): a learning process enriched by gamification but also close to the learner’s activities and emotional setting.
After a brief introduction on related work, we describe the technological setup, the integration of projected visual feedback and the use of object and motion recognition to interpret the learner’s actions. We explain how this combination enables rapid feedback and why this is particularly important for correct habit formation in practical skills training. In a second step, we discuss gamification methods and analyze which are best suited for the PCL system. Finally, emotion recognition, a major element of the final PCL design not yet implemented, is briefly outlined.

Universal Kinect-type-controller by ICE Lab (UKI, pronounced as “You-key”) is developed for allowing users to control any existing applications by using body motions as inputs. The middleware works by converting detected motions into keyboard and/or mouse-click events, and sending them to a target application. This paper presents the structures and designs of core modules, along with examples from real cases to illustrate how the middleware can be configured to fit a variety of applications. We present our designs of interfaces that decode all configuration details into a human-interpretable language, and these interfaces significantly promote user experience and eliminate user divide in, for example, programming skill. The performance of the middleware is evaluated on fighting-game motion data, and we make these data publicly available so that they can be used in other researches. UKI welcomes its use to everyone in unlimited purpose; for instance, it can be used to promote healthy life through a means of gaming and/or used to conduct serious researches on motion systems. The middleware serves as a shortcut in the development of motion applications – coding of an application to detect motions can be replaced with simple clicks on UKI.

Australian and Singaporean students have been exposed to different forms of teaching due to cultural differences in education. In each country, varying degrees of importance has been placed on explicitly teaching problem-solving heuristics. This paper highlights the different strategies employed by students from each country when solving a word problem and the role direct teaching has played in the development of these strategies. Implications for classroom practice are also discussed.

This paper presents an approach towards developing a universal Kinect interface supporting game-playing. The proposed interface is designed to be used with existing fighting games of any sort. Players can define their own postures and map them to keystrokes for sending input to a targeted game without the need of accessing game source code. We describe the overview of the interface system; including motion segmentation, posture feature extraction, and posture-to-key mapping.

This paper presents a full-body motion-control game interface based on a Kinect device. A set of postures and motions for controlling game characters is presented, and the posture-detection algorithm for each posture is implemented by using a rule-based technique with rule-and-threshold optimization. Our experiments show that the proposed optimization can improve the accuracy of posture detection by 13.70%. The proposed techniques are applied to the fighting game called FightingICE, a game platform in recent CIG competitions.

This study examines the efficiency of human
motion-based UI for video games with motion
capture system, Kinect. We took an investigation to
play with the Kinect sensor in the running game
which was developed and designed using two kinds
of UI. One UI consists of more intuitive and
familiar motions such as turning and jumping. The
other UI consists of arm motions like raising hands.
As a result, UI with arm motions was easier for
users to master and results in higher success rates to
play than the other UI. Therefore we can conclude
when a game is developed using Kinect and its UI is
configured with motion recognition, the motion
with the arms rather than the other parts of the body
helps player better to enhance the play skills and
immerse in the game.

Virtual reality therapy has been successfully applied into visualization area. However, they can sometimes leave the user in a virtual world that is not real. Augmented Reality (AR) on the other hand is able to utilize the advantage of having virtual graphical elements that are superimposed onto the real world scene. In this paper, we propose a semi-portable system for upper limb disability. This system uses the Microsoft Kinect™ together with two wireless sensors being strapped on the upper limbs to track the movement of the arm of the patients. Together with AR, the patient will be able to perform a set of targeted exercises.