• Journal of KIISE
• /
• v.42 no.3
• /
• pp.295-306
• /
• 2015

MOnCa2 is a framework for building intelligent smartphone applications based on smartphone sensors and ontology reasoning. In previous studies, MOnCa determined and inferred user situations based on sensor values represented by ontology instances. When this approach is applied, recognizing user space information or objects in user surroundings is possible, whereas determining the user's physical context (travel behavior, travel destination) is impossible. In this paper, MOnCa2 is used to build recognition models for travel behavior and routes using smartphone sensors to analyze the user's physical context, infer basic context regarding the user's travel behavior and routes by adapting these models, and generate high-level context by applying ontology reasoning to the basic context for creating intelligent applications. This paper is focused on approaches that are able to recognize the user's travel behavior using smartphone accelerometers, predict personal routes and destinations using GPS signals, and infer high-level context by applying realization.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.04a
• /
• pp.340-342
• /
• 2015

스마트폰에 장착된 센서들을 사용하면 다양한 물리량을 측정할 수 있다. 본 논문에서는 스마트폰을 사용하여 물체의 높이를 추정하는 기법을 제시한다. 제안 기법은 물체를 여러 위치에서 관측하여 그 높이를 추정하는 것이다. 각각의 관측점에서는 스마트폰의 위치센서와 방향센서를 사용하여 관측점의 위치와 관측 방향을 측정한다. 이러한 관측값을 기반으로 각 관측점에서 물체로 향하는 3차원 시선백터를 도출하고 그것들 간의 최근점점을 계산하여 물체의 높이를 추정한다. 제안 기법을 모바일앱으로 구현하고 건축물 높이를 측정하는 실험을 통해 성능을 시험하였다.

• Journal of rehabilitation welfare engineering & assistive technology
• /
• v.9 no.3
• /
• pp.245-250
• /
• 2015

Portable devices, such as smart phones and personal digital assistants (PDAs) play an important role in our everyday life. In this paper, we propose a step count algorithm based on SVM(signal vector magnitude) and a adaptive threshold processing to monitor the physical activity. The algorithm measures a user's step counts using the smart phone's inbuilt accelerometer and g sensor. Experiment results showed the proposed algorithm has good performance in accuracy and adaptability than the app on your smart phone.

• Smart Media Journal
• /
• v.12 no.3
• /
• pp.9-18
• /
• 2023

Due to the deepening of COVID-19, high-intensity social distancing has been prolonged and many social problems have been cured. In particular, physical and psychological isolation occurred due to the non-face-to-face system and a lot of damage occurred. The various social problems caused by Corona acted as severe stress for all those affected by Corona 19, and eventually acted as a factor threatening mental health such as depression. While the number of people suffering from mental illness is increasing, the actual use of mental health services is low. Therefore, it is necessary to establish a system for people suffering from mental health problems. Therefore, in this study, depression detection and emergency detection models were constructed based on sensor information using smartphones from depressed subjects and general subjects. For the detection of depression and emergencies, VAE, DAGMM, ECOD, COPOD, and LGBM algorithms were used. As a result of the study, the depression detection model had an F1 score of 0.93 and the emergency situation detection model had an F1 score of 0.99. direction.

• Proceedings of the Korean Information Science Society Conference
• /
• 2011.06d
• /
• pp.38-40
• /
• 2011

최근 상황인지에 관한 연구가 활발히 진행되고 있다. 스마트폰의 각종 센서를 통해 사용자의 컨텍스트 수집이 가능해졌고 이러한 사용자의 컨텍스트는 사용자에게 보다 친화적인 서비스를 제공하기 위한 데이터로 활용이 가능하다. 컨텍스트는 물리적 컨텍스트(Physical Context)와 소프트 컨텍스트(Soft Context)로 구분할 수 있는데 이 두 가지의 컨텍스트를 조합하면 사용자의 취향과 상황 그리고 생활 패턴 등을 보다 정확하게 파악할 수 있다. 이렇게 상황정보를 이용하여 추출된 데이터는 사용자에게 친화적인 서비스를 제공할 수 있는 토대로 활용할 수 있다. 본 논문에서는 사용자의 상황 정보에 기반을 둔 로그 수집 방법과 분석방법을 제시하여 사용자의 상황에 적합한 애플리케이션을 추천하는 시스템을 설계하고 구현하였다. 애플리케이션 추천 시스템은 소프트 컨텍스트와 물리적 컨텍스트의 조합으로 생성한 통계정보를 사용하기 때문에 보다 사용자에게 친화적으로 애플리케이션을 추천할 수 있다. 또한 애플리케이션 추천 시스템은 애플리케이션 카테고리 또는 애플리케이션 사용 횟수에 따른 분류 등으로 사용자의 스마트폰 활용패턴을 통계정보로 나타내준다. 애플리케이션 추천 시스템을 사용함으로써 사용자는 개인에게 가장 알맞은 스마트폰 환경을 사용할 수 있으며, 자신의 애플리케이션 활용 패턴 및 통계정보도 숙지할 수 있어 사용자에게 보다 밀접한 스마트폰 활용 정보를 제공할 수 있다. 이러한 상황정보 기반의 로그 분석과 수집, 그리고 애플리케이션 추천 시스템은 추후 상황인지 및 사용자의 특화된 서비스 개발에 많은 도움이 될 것이다.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.

• Information and Communications Magazine
• /
• v.30 no.8
• /
• pp.3-10
• /
• 2013

사물인터넷(IoT: Internet of Things)은 지능화된 사물들이 연결되는 네트워크를 통해 사람과 사물 (물리 또는 가상), 사물과 사물간에 상호 소통하고 상황인식 기반의 지식이 결합되어 지능적인 서비스를 제공하는 글로벌 인프라이며, 2017년 2,900억 달러 시장이 예측되는 스마트 폰 이후 유망기술이며 모바일, 클라우드, 빅데이터 기술 등과 융합하여 초연결사회의 핵심이 될 전망이다. IoT의 성공적인 실현을 위해서는 데이터 분석 및 추론, 개방형 시맨틱 플랫폼, 고신뢰 네트워크, 센서-스마트 단말 인터렉션 및 협업, 에너지 하베스팅, 스마트센서 등의 핵심기술 개발과 글로벌 표준화, 정보보호, 사생활 침해 우려 등의 장애 극복을 위해 IoT 생태계 참여자 모두의 협력이 필요하다.

• Journal of Korea Multimedia Society
• /
• v.15 no.8
• /
• pp.1059-1066
• /
• 2012

With the development of Smartphone, Smartphone contains diverse functions including many sensors that can describe users' state. So there has been increased studies rapidly about activity recognition and life pattern recognition with Smartphone sensors. This research suggest modeling of the activity data to classify extracted data in existing activity recognition study. Activity data is divided into two parts: Physical activity and Logical Activity. In this paper, activity data modeling is theoretical analysis. We classified the basic activity(walking, standing, sitting, lying) as physical activity and the other activities including object, target and place as logical activity. After that we suggested a method of visualizing modeling data for users. Our approach will contribute to generalize human's life by modeling activity data. Also it can contribute to visualize user's activity data for existing activity recognition study.

• Journal of Korea Game Society
• /
• v.14 no.2
• /
• pp.67-76
• /
• 2014

Recently there have been many interests on tangible sport games that can recognize the motions of players. In this paper, we propose essential technologies required for tangible games, which are methods for swing motion recognition and the calculation of shuttle cock's trajectory. When a user carries out a badminton swing while holding a smartphone with his hand, the motion signal generated by smartphone-embedded acceleration sensors is transformed into a feature vector through a Daubechies filter, and then its swing type is recognized using a k-NN based method. The method for swing motion presented herein provides an advantage in a way that a player can enjoy tangible games without purchasing a commercial motion controller. Since a badminton shuttle cock has a particular flight trajectory due to the nature of its shape, it is not easy to calculate the trajectory of the shuttle cock using simple physics rules about force and velocity. In this paper, we propose a method for calculating the flight trajectory of a badminton shuttle cock in which the wind effect is considered.

• Proceedings of the Korean Information Science Society Conference
• /
• 2012.06d
• /
• pp.149-151
• /
• 2012

최근 상황인지에 관한 연구가 활발히 진행되고 있다. 스마트폰의 각종 센서를 통해 사용자의 컨텍스트 수집이 가능해졌고 이러한 사용자의 컨텍스트는 사용자에게 보다 친화적인 서비스를 제공하기 위한 데이터로 활용이 가능하다. 컨텍스트는 물리적 컨텍스트(Physical Context)와 소프트 컨텍스트(Soft Context)로 구분할 수 있다. 이렇게 상황정보를 이용하여 추출된 데이터는 사용자에게 친화적인 서비스를 제공할 수 있는 토대로 활용할 수 있다. 하지만 물리적 컨텍스트만을 이용하는 기존의 방법은 실제로 동적인 사용자의 컨텍스트를 정확하게 유추하기 어려운 구조이다. 본 논문에서는 모바일 기기에서 사용자에게 보다 친화적인 서비스를 제공하기 위해서 소프트 컨텍스트를 사용하여 Logical Sensor를 설계 및 구현한다. 여기서 Logical Sensor는 소프트 컨텍스트를 통해서 사용자의 소셜 네트워크나 사용자의 선호도를 파악하여 로그데이터를 남긴다. 이렇게 얻은 로그데이터는 통계를 통해 사용자의 선호도나 소셜 네트워크를 한눈에 볼 수 있으며, 시간이나 위치에 따라서 사용자가 모바일에서 사용할 애플리케이션이나 통화상대등을 추천 해줄 수 있을 것이다. 이뿐만 아니라 Logical Sensor로부터 얻은 사용자의 로그 데이터는 사용자에게 사용자의 특화된 서비스 개발에 많은 도움이 될 것이다.