• The KIPS Transactions:PartB
• /
• v.18B no.1
• /
• pp.11-20
• /
• 2011

Pictorial Structures is known as an effective method that recognizes and tracks human poses. In this paper, the upper body pose is also tracked by PS and a particle filter(PF). PF is one of dynamic programming methods. But Markov chain-based dynamic motion model which is used in dynamic programming methods such as PF, couldn't predict effectively the highly articulated upper body motions. Therefore PF often fails to track upper body pose. In this paper we propose the key pose-based proposal distribution for proper particle prediction based on the similarities between key poses and an upper body silhouette. In the experimental results we confirmed our 70.51% improved performance comparing with a conventional method.

• Proceedings of the Korean Information Science Society Conference
• /
• 2007.06b
• /
• pp.176-181
• /
• 2007

인체와 같은 복잡한 계층 구조를 가진 관절체에서 모션을 생성하는 것은 쉬운 일이 아니다. 기존의 캐릭터 포즈 에디팅 방법은 정운동학과 역운동학을 사용하여 키프레임을 생성하였다. 본 논문에서는 역운동학에 기반 한 직관적인 포즈 편집 기법을 제안한다. 선택된 연속적인 관절들을 입력으로 주어진 커브에 정렬시킴으로써 원하는 포즈를 쉽게 생성할 수 있다. 이렇게 만들어진 포즈는 키 프레임 애니메이션의 키프레임으로 사용될 수 있다. 본 논문에서는 커브에 관절의 연속된 부분을 정렬시키기 위해 점진적 관절 정렬 기법(Gradual Joint Alignment along Curve)을 제시하고 이 방법을 통해 연속된 조인트 체인으로 하여금 역운동학의 제약 조건을 만족시키면서도 가능한 한 입력된 커브에 정렬될 수 있도록 하였다.

• Cartoon and Animation Studies
• /
• s.41
• /
• pp.57-73
• /
• 2015

A pose is an important component in the animation with timing and spacing. Pose is the key to describe the story-telling or how the animation behavior. Key animation method is Straight Ahead and pose to pose method. Many animaters have been using these two methods, or by a mix of two ways. It is possible that computer animation make a pose using interpolation between keyframes. The many animators of computer animation are using pose to pose in their work. It is depend on good and strong pose that make audience understand a story or a situation. This makes animators to be efficient of inefficient operation. In this study, according to the effective good pose to catch proposes four ways. There are four methods of making pose that are stretch and squash, the height of the character, the center of weight, step. The law of 12 kinds of Disney Animation is a good reference for the study.

• Journal of KIISE:Software and Applications
• /
• v.35 no.12
• /
• pp.780-788
• /
• 2008

Modeling hand poses and tracking its movement are one of the challenging problems in computer vision. There are two typical approaches for the reconstruction of hand poses in 3D, depending on the number of cameras from which images are captured. One is to capture images from multiple cameras or a stereo camera. The other is to capture images from a single camera. The former approach is relatively limited, because of the environmental constraints for setting up multiple cameras. In this paper we propose a method of reconstructing 3D hand poses from a 2D input image sequence captured from a single camera by means of Belief Propagation in a graphical model and recognizing a finger clicking motion using a hidden Markov model. We define a graphical model with hidden nodes representing joints of a hand, and observable nodes with the features extracted from a 2D input image sequence. To track hand poses in 3D, we use a Belief Propagation algorithm, which provides a robust and unified framework for inference in a graphical model. From the estimated 3D hand pose we extract the information for each finger's motion, which is then fed into a hidden Markov model. To recognize natural finger actions, we consider the movements of all the fingers to recognize a single finger's action. We applied the proposed method to a virtual keypad system and the result showed a high recognition rate of 94.66% with 300 test data.

• Smart Media Journal
• /
• v.12 no.7
• /
• pp.59-67
• /
• 2023

Recently, research on behavior analysis tracking human posture and movement has been actively conducted. In particular, OpenPose, an open-source software developed by CMU in 2017, is a representative method for estimating human appearance and behavior. OpenPose can detect and estimate various body parts of a person, such as height, face, and hands in real-time, making it applicable to various fields such as smart healthcare, exercise training, security systems, and medical fields. In this paper, we propose a method for classifying four exercise movements - Squat, Walk, Wave, and Fall-down - which are most commonly performed by users in the gym, using OpenPose-based deep learning models, DNN and CNN. The training data is collected by capturing the user's movements through recorded videos and real-time camera captures. The collected dataset undergoes preprocessing using OpenPose. The preprocessed dataset is then used to train the proposed DNN and CNN models for exercise movement classification. The performance errors of the proposed models are evaluated using MSE, RMSE, and MAE. The performance evaluation results showed that the proposed DNN model outperformed the proposed CNN model.

• Proceedings of the Korean Society for Emotion and Sensibility Conference
• /
• 2008.10a
• /
• pp.131-134
• /
• 2008

오늘날 3D 캐릭터 애니메이션은 실사영화, 애니메이션, 게임, 광고 등 대다수의 영상물에서 쉽게 접할 수 있다. 캐릭터의 부드러운 움직임은 모션캡쳐(Motion Capture)나 숙련된 애니메이터의 키 프레임(Key Frame) 작업의 결과물일 것이다. 이런 작업들은 고가의 장비나 많은 인력을 요구하고 완성된 결과물은 수정하거나 효과를 주기가 힘들다. 본 연구에서는 3D Studio MAX Script를 이용한 삼차원 회전 값의 연산으로 바이페드(Biped)의 포즈나 애니메이션을 합성하고 보다 사실적인 합성을 위한 방법을 제시하고자 한다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2014.11a
• /
• pp.88-89
• /
• 2014

본 논문에서는 미리 모델링된 실객체에 대한 3차원 모델을 변형하면서 애니메이션을 정의하고, 실사 객체를 추적하면서 카메라 뷰 상의 실사 객체에 애니메이션을 적용하는 방법을 제안한다. 애니메이션 정의는 라플라시안 기반 메쉬 변형 방법으로 3차원 모델을 변형시키며 키프레임을 지정함으로써 이루어진다. 정의된 애니메이션은 실제 환경에서 추적된 실사 객체의 영상으로부터 모델에 텍스처를 입힌 뒤, 카메라 포즈를 이용하여 객체 위치에 객체 모델을 렌더링할 때 적용된다. 제안된 방법을 통해 사용자가 원하는 대로 실제 환경에 대한 카메라 뷰 상에서 실사 객체가 변형되는 모습을 용이하게 보여줄 수 있다.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.19 no.1
• /
• pp.217-224
• /
• 2024

With the rapid growth of K-POP, the dance content industry is spreading. With the recent increase in the spread of SNS, they also shoot and share their dance videos. However, it is not easy for dance beginners who are new to dancing to learn dance moves because it is difficult to receive objective feedback when dancing alone while watching videos. This paper describes a system that uses MediaPipe to compare choreography videos and dance videos of users and detect whether they are following the movement correctly. This study proposes a method of giving feedback based on Color Map to users by calculating the similarity of dance movements between user images taken with webcam or camera and choreography images using cosine similarity and COCO OKS. Through this system, objective feedback on users' dance movements can be visually received, and beginners are expected to be able to learn accurate dance movements.