http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
강창구(Changgu Kang),이석원(Sukwon Lee) 한국정보기술학회 2022 한국정보기술학회논문지 Vol.20 No.11
In everyday life, people continue to live with various physical interactions with the environment, such as walking, opening the door, drinking water, talking on the phone, and sitting down. Various studies on such interactions are being conducted. The physical interaction starts with contact, and interaction operations may be classified according to the contact portion of the person. Despite this importance, most studies have focused on interactive contact posture or behavioral induction, and studies on determining whether or not to contact a physical object have been lacking. To overcome this situation, we propose a method for detecting hand contact states where the most frequent contact occurs in physical interactions. The proposed method consists of a bidirectional LSTM based feature vector generation part and a contact detection part. In addition, the contact detection part has a learning structure in which the left and right hands are divided. We perform model learning using the collected motion data, and the test results of the learned model shows 93.9% accuracy.
강창구(Changgu Kang),김미경(Meekyoung Kim) 한국정보기술학회 2022 한국정보기술학회논문지 Vol.20 No.10
Research on the analysis of interaction behavior in the environment is being conducted in various fields. Interaction begins with the act of contact with the environment, and contact is recognized as important contextual information. Despite this importance, accurate contact recognition is difficult due to the complexity of the environment and the occlusion of people or objects, and most studies require environment data for contact detection. In this paper, we propose a foot contact detection method, which is important contact information for accurate behavior recognition, overcoming these constraints. The proposed method utilizes a recursive neural network suitable for the continuity of motion data, and uses the spatial and temporal information processed by the motion data as a feature for learning. Finally, we perform model training using the collected motion data, and the test result of the trained model shows about 95% accuracy.
긍정감정을 유도하기 위한 모방학습을 이용한 상호작용 시스템 프로토타입 개발
오찬해(Chanhae Oh),강창구(Changgu Kang) 한국정보전자통신기술학회 2021 한국정보전자통신기술학회논문지 Vol.14 No.4
컴퓨터 그래픽스 및 HCI 분야에서 캐릭터를 만들고 자연스럽게 상호작용하는 시스템에 관한 많은 연구가 있었다. 이와 같은 연구들은 사용자의 행동에 대한 반응에 중점을 두었으며, 사용자에게 긍정적 감정을 끌어내기 위한 캐릭터의 행동 연구는 여전히 어려운 문제로 남아있다. 본 논문에서는 인공지능 기술을 이용하여 가상 캐릭터의 움직임에 따른 사용자의 긍정적 감정을 끌어내기 위한 상호작용 시스템 프로토타입을 개발한다. 제안된 시스템은 표정 인식과 가상 캐릭터의 동작 생성으로 구분된다. 표정 인식을 위해 깊이 카메라를 사용하며 인식된 사용자의 표정 데이터는 동작 생성으로 전달된다. 우리는 개인화된 상호작용 시스템 개발을 위하여 학습모델로서 모방학습을 사용한다. 동작 생성에서는 최초 사용자의 표정 데이터에 따라 무작위 행동을 수행하고 지속적인 모방학습을 통하여 사용자가 긍정적 감정을 끌어낼 수 있는 행동을 학습한다. In the field of computer graphics and HCI, there are many studies on systems that create characters and interact naturally. Such studies have focused on the user"s response to the user"s behavior, and the study of the character"s behavior to elicit positive emotions from the user remains a difficult problem. In this paper, we develop a prototype of an interaction system to elicit positive emotions from users according to the movement of virtual characters using artificial intelligence technology. The proposed system is divided into face recognition and motion generation of a virtual character. A depth camera is used for face recognition, and the recognized data is transferred to motion generation. We use imitation learning as a learning model. In motion generation, random actions are performed according to the first user"s facial expression data, and actions that the user can elicit positive emotions are learned through continuous imitation learning.
현실환경과 가상 환경을 연동하는 맥락 인식 기반 U-VR 시뮬레이터
오유수(Yoosoo Oh),강창구(Changgu Kang),우운택(Woontack Woo) 한국HCI학회 2009 한국HCI학회 학술대회 Vol.2009 No.2
본 논문에서는 유비쿼터스 가상 현실 개념에 기반하여 맥락 정보를 이용하여 현실과 가상 환경을 이음매 없이 연결하는 U-VR 시뮬레이터를 제안한다. U-VR 시뮬레이터는 스마트 홈 환경을 시뮬레이션 할 수 있고, 센서, 액추에이터, 서비스와 같은 새로운 가상 개체들을 기존의 실제 개체에 추가할 수 있게 한다. 제안된 시뮬레이터는 기존의 장치와 시뮬레이션 할 장치를 동시에 활용함으로써 빠른 개발과 비용 절감의 효과가 있다. 또한, 제안된 시뮬레이터는 어플리케이션 개발자들이 유비쿼터스 가상 현실 응용을 쉽고 빠르게 개발하고, 적절한 도메인으로 확장하게 한다. 추후 연구로는 제안된 시뮬레이터가 현실 환경에 가상의 개체들을 지능적으로 증강하도록 확장할 것이다. In this paper we propose the U-VR Simulator which seamlessly connects entities in real and virtual environments by exploiting context, based on the concept of ubiquitous virtual reality. Using the U-VR Simulator, a smart home environment can be simulated and new virtual entities such as sensors, actuators, and services can be added to existing real entities. The proposed approach is fast and cost-effective because the simulator effectively utilizes both existing and simulated devices. In addition, our approach allows application developers to rapidly develop U-VR applications and extend them to relevant domains. As a future work, we will expand our simulator to intelligently augment virtual entities into real environments.
임원주(Won-Ju Im),황인태(In-Tae Hwang),강창구(Changgu Kang) 한국정보기술학회 2023 한국정보기술학회논문지 Vol.21 No.3
Technological advances on the 4th industrial revolution are bringing about many changes in various fields. In particular, in the field of education, it has had positive effects such as inducing students’ interest in learning and inducing creativity and active participation. In this paper, we propose an educational game for arithmetic operations using augmented reality technology, one of these technologies. In the proposed educational game, 3D contents related to 4 arithmetic operations using a display device are augmented, and physical interaction with learners using a merge cube is provided. In addition, the continuous participation of the learner is induced through compensation according to the result of the given problem. Lastly, we expect learners to directly manipulate the merge cube and play educational games to provide interest in learning, enable long-term memory, and help improve learners’ logical skills through the correct combination of operators in a given expression.