• 인터넷정보학회논문지
• /
• 제21권5호
• /
• pp.21-29
• /
• 2020

Accurate estimation of human pose relies on backbone method in which its role is to extract feature map. Up to dated, the method of backbone feature extraction is conducted by the plain convolutional neural networks named by CNN and the residual neural networks named by Resnet, both of which have various architectures and performances. The CNN family network such as VGG which is well-known as a multiple stacked hidden layers architecture of deep learning methods, is base and simple while Resnet which is a bottleneck layers architecture yields fewer parameters and outperform. They have achieved inspired results as a backbone network in human pose estimation. However, they were used then followed by different pose estimation networks named by pose parsing module. Therefore, in this paper, we present a comparison between the plain CNN family network (VGG) and bottleneck network (Resnet) as a backbone method in the same pose parsing module. We investigate their performances such as number of parameters, loss score, precision and recall. We experiment them in the bottom-up method of human pose estimation system by adapted the pose parsing module of openpose. Our experimental results show that the backbone method using VGG network outperforms the Resent network with fewer parameter, lower loss score and higher accuracy of precision and recall.

• 한국재난정보학회 논문집
• /
• 제20권2호
• /
• pp.302-314
• /
• 2024

연구목적: 본 연구는 지하공동구 내 다수 작업자의 낙상을 자동으로 판별하기 위한 Top-down 방식의 딥러닝 자세 추정 모델 기반 낙상 검출 모델을 제안하고, 제안 모델의 성능을 평가한다. 연구방법: Top-down 방식의 자세 추정모델 중 하나인 YOLOv8-pose로부터 추론된 결과와 낙상 판별 규칙을 결합한 모델을 제시하고, 지하공동구 내 2인 이하 작업자가 출현한 기립 및 낙상 이미지에 대해 모델 성능지표를 평가하였다. 또한 동일한 방법으로 Bottom-up 방식 자세추정모델(OpenPose)을 적용한 결과를 함께 분석하였다. 두 모델의 낙상 검출 결과는 각 딥러닝 모델의 작업자 인식 성능에 의존적이므로, 작업자 쓰러짐과 함께 작업자 존재 여부에 대한 성능지표도 함께 조사하였다. 연구결과: YOLOv8-pose와 OpenPose의 모델의 작업자 인식 성능은 F1-score 기준으로 각각 0.88, 0.71로 두 모델이 유사한 수준이었으나, 낙상 규칙을 적용함에 따라 0.71, 0.23로 저하되었다. 작업자의 신체 일부만 검출되거나 작업자간 구분을 실패하여, OpenPose 기반 낙상 추론 모델의 성능 저하를 야기한 것으로 분석된다. 결론: Top-down 방식의 딥러닝 자세 추정 모델을 사용하는 것이 신체 관절점 인식 및 개별 작업자 구분 측면에서 지하공동구 내 작업자 낙상 검출에 효과적이라 판단된다.

• 대한임베디드공학회논문지
• /
• 제11권3호
• /
• pp.183-191
• /
• 2016

Recently, in rehabilitation treatment, the situation that requires a measure of the accuracy of the pose and movement of joints is being increased due to the habits and lifestyle of modern people and the environment. In particular, there is a need for active automated system that can determine itself for the matching rate of pose Basically, a method for measuring the matching rate of pose is used by extracting an image using the Kinect or extracting a silhouette using the imaging device. However, in the case of extracting a silhouette, it is difficult to set the comparison, and in the case of using the Kinect sensor, there is a disadvantages that high accumulated error rate according to movement. Therefore, In this paper, we propose a method to reduce the accumulated error of matching rate of pose getting the rotation angle of joint by measuring the real-time amount of change of 9-axis sensor. In particular, it can be measured same conditions that unrelated of the physical condition and unaffected by the data for the back and forth movement, because of it compares the current rotation angle of the joint. Finally, we show a comparative advantage results by compared with traditional method of extracting a silhouette and a method using a Kinect sensor.

• Journal of information and communication convergence engineering
• /
• 제21권4호
• /
• pp.337-345
• /
• 2023

In this study, we formulated a method that evaluates Taekwondo Poomsae performance using a series of choreographed training movements. Despite recent achievements in 3D human pose estimation (HPE) performance, the analysis of human actions remains challenging. In particular, Taekwondo Poomsae action analysis is challenging owing to the absence of time synchronization data and necessity to compare postures, rather than directly relying on joint locations owing to differences in human shapes. To address these challenges, we first decomposed human joint representation into joint rotation (posture) and limb length (body shape), then synchronized a comparison between test and reference pose sequences using DTW (dynamic time warping), and finally compared pose angles for each joint. Experimental results demonstrate that our method successfully synchronizes test action sequences with the reference sequence and reflects a considerable gap in performance between practitioners and professionals. Thus, our method can detect incorrect poses and help practitioners improve accuracy, balance, and speed of movement.

• 스마트미디어저널
• /
• 제12권7호
• /
• pp.59-67
• /
• 2023

최근 인간의 자세와 행동을 추적하는 행동 분석 연구가 활발해지고 있다. 특히 2017년 CMU에서 개발한 오픈소스인 오픈포즈(OpenPose)는 사람의 외모와 행동을 추정하는 대표적인 방법이다. 오픈포즈는 사람의 키, 얼굴, 손 등의 신체부위를 실시간으로 감지하고 추정할 수 있어 스마트 헬스케어, 운 동 트레이닝, 보안시스템, 의료 등 다양한 분야에 적용될 수 있다. 본 논문에서는 헬스장에서 사용자들이 가장 많이 운동하는 Squat, Walk, Wave, Fall-down 4개 동작을 오픈포즈기반 딥러닝인 DNN과 CNN을 이용하여 운동 동작 분류 방법을 제안한다. 학습데이터는 녹화영상 및 실시간으로 카메라를 통해 사용자의 동작을 캡처해서 데이터 셋을 수집한다. 수집된 데이터 셋은 OpenPose을 이용하여 전처리과정을 진행하고, 전처리과정이 완료된 데이터 셋은 본 논문에서 제안한 DNN 및 CNN 모델 이용하여 운동 동작 분류를 학습한다. 제안한 모델에 대한 성능 오차는 MSE, RMSE, MAE를 사용한다. 성능 평가 결과, 제안한 DNN 모델 성능이 제안한 CNN 모델보다 우수한 것으로 나타났다.

• 한국컴퓨터정보학회논문지
• /
• 제23권8호
• /
• pp.9-16
• /
• 2018

In this paper, a kidnapping detection scheme in which human pose estimation is used to classify accurately between kidnapping cases and normal ones is proposed. To estimate human poses from input video, human's 10 joint information is extracted by OpenPose library. In addition to the features which are used in the previous study to represent the size change rates and the regularities of human activities, the human pose estimation features which are computed from the location of detected human's joints are used as the features to distinguish kidnapping situations from the normal accompanying ones. A frame-based kidnapping detection scheme is generated according to the selection of J48 decision tree model from the comparison of several representative classification models. When a video has more frames of kidnapping situation than the threshold ratio after two people meet in the video, the proposed scheme detects and notifies the occurrence of kidnapping event. To check the feasibility of the proposed scheme, the detection accuracy of our newly proposed scheme is compared with that of the previous scheme. According to the experiment results, the proposed scheme could detect kidnapping situations more 4.73% correctly than the previous scheme.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2005년도 ICCAS
• /
• pp.735-739
• /
• 2005

This paper presents a new system to estimate the head pose of human in interactive indoor environment that has dynamic illumination change and large working space. The main idea of this system is to suggest a new morphological feature for estimating head angle from stereo disparity map. When a disparity map is obtained from stereo camera, the matching confidence value can be derived by measurements of correlation of the stereo images. Applying a threshold to the confidence value, we also obtain the specific morphology of the disparity map. Therefore, we can obtain the morphological shape of disparity map. Through the analysis of this morphological property, the head pose can be estimated. It is simple and fast algorithm in comparison with other algorithm which apply facial template, 2D, 3D models and optical flow method. Our system can automatically segment and estimate head pose in a wide range of head motion without manual initialization like other optical flow system. As the result of experiments, we obtained the reliable head orientation data under the real-time performance.

• 한국CDE학회논문집
• /
• 제16권4호
• /
• pp.285-291
• /
• 2011

A pose estimation process from medical images is calculating locations and orientations of objects obtained from Computed Tomography (CT) volume data utilizing X-ray images from two directions. In this process, digitally reconstructed radiograph (DRR) images of spatially transformed objects are generated and compared to X-ray images repeatedly until reasonable transformation matrices of the objects are found. The DRR generation and image comparison take majority of the total time for this pose estimation. In this paper, a fast DRR generation technique based on GPU parallel computing is introduced. A volume ray-casting algorithm is explained with brief vector operations and a parallelization technique of the algorithm using Compute Unified Device Architecture (CUDA) is discussed. This paper also presents the implementation results and time measurements comparing to those from pure-CPU implementation and open source toolkit.

• Journal of Computing Science and Engineering
• /
• 제10권3호
• /
• pp.95-101
• /
• 2016

In this study, we propose a new system for constructing parallel corpora for sign languages, which are generally under-resourced in comparison to spoken languages. In order to achieve scalability and accessibility regarding data collection and corpus construction, our system utilizes deep learning-based techniques and predicts depth information to perform pose estimation on hand information obtainable from video recordings by a single RGB camera. These estimated poses are then transcribed into expressions in SignWriting. We evaluate the accuracy of hand tracking and hand pose estimation modules of our system quantitatively, using the American Sign Language Image Dataset and the American Sign Language Lexicon Video Dataset. The evaluation results show that our transcription system has a high potential to be successfully employed in constructing a sizable sign language corpus using various types of video resources.

• Smart Structures and Systems
• /
• 제14권5호
• /
• pp.943-960
• /
• 2014

A visually servoed paired structured light system (ViSP) was recently proposed as a novel estimation method of the 6-DOF (Degree-Of-Freedom) relative displacement in civil structures. In order to apply the ViSP to massive structures, multiple ViSP modules should be installed in a cascaded manner. In this configuration, the estimation errors are propagated through the ViSP modules. In order to resolve this problem, a displacement estimation error back-propagation (DEEP) method was proposed. However, the DEEP method has some disadvantages: the displacement range of each ViSP module must be constrained and displacement errors are corrected sequentially, and thus the entire estimation errors are not considered concurrently. To address this problem, a pose-graph optimized displacement estimation (PODE) method is proposed in this paper. The PODE method is based on a graph-based optimization technique that considers entire errors at the same time. Moreover, this method does not require any constraints on the movement of the ViSP modules. Simulations and experiments are conducted to validate the performance of the proposed method. The results show that the PODE method reduces the propagation errors in comparison with a previous work.