• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.461-463
• /
• 2023

본 연구는 실시간 MOT(Multiple-Object-Tracking)의 성능을 향상시키기 위해 다양한 기법을 적용한 MOT 모델을 설계한다. 연구에서 사용하는 Backbone 모델은 TBD(Tracking-by-Detection) 기반의 Tracking 모델을 사용한다. Heatmap Detection을 통해 객체를 검출하고 Transformer 기반의 Feature를 연결하여 Tracking 한다. 제안하는 방법은 Anchor 기반의 Detection의 장시간 문제와 추적 객체 정보 전달손실을 감소하여 실시간 객체 추적에 도움이 될 것으로 사료된다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.539-540
• /
• 2023

최근 물체 인식 모델의 성능을 개선하기 위한 다양한 연구가 진행 중이다. 본 논문에서는 K-means 기반 앵커박스 선정 기법을 적용한 새로운 물체 인식 모델 성능 개선 방법을 제안한다. 제안된 방법은 항만 내 설치된 컨테이너 사고를 예방하기 위한 컨테이너 사고위험도 분류 모델에 적용하여 성능 평가를 하였다. 특히, 컨테이너 사고위험도 분류 모델은 작은 물체를 인식해야 하며 이런 환경에서는 기존 물체 인식 모델 성능이 낮게 나타난다. 본 논문에서는 제안한 K-means 기반 앵커박스 선정 기법을 적용하여 물체 인식 모델 성능이 개선됨을 확인하였디.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.509-511
• /
• 2023

현대 컴퓨터 비전 분야에서는 객체 추적이 중요한 연구 주제 중 하나다. 기존 Tracking By Detection 방식은 실시간 추적 속도와 Tracklet을 유지할 수 있는 정보 전달의 한계를 가지고 있다. 본 연구에서는 유사도 비교 모듈을 기반으로 Tracking By Detection 모델을 설계하고자 한다. 탐지 모델은 Anchor를 사용하지 않는 CenterNet을 사용하고 탐지된 값에 유사도 비교 알고리즘을 적용하여 객체 탐지와 객체 추적을 동시에 수행하는 모델을 제안한다. 제안하는 방법은 Occlusion으로 인한 객체 정보 손실을 완화하고, 새로운 객체 및 장애물에 대해 강건할 것으로 사료된다.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.5
• /
• pp.225-231
• /
• 2023

With the recent surge in the autonomous driving market, the significance of lane detection technology has escalated. Lane detection plays a pivotal role in autonomous driving systems by identifying lanes to ensure safe vehicle operation. Traditional lane detection models rely on engineers manually extracting lane features from predefined environments. However, real-world road conditions present diverse challenges, hampering the engineers' ability to extract adaptable lane features, resulting in limited performance. Consequently, recent research has focused on developing deep learning based lane detection models to extract lane features directly from data. In this paper, we classify lane detection models into four categories: cluster-based, curve-based, information propagation-based, and anchor-based methods. We conduct an extensive analysis of the strengths and weaknesses of each approach, evaluate the model's performance on an embedded board, and assess their practicality and effectiveness. Based on our findings, we propose future research directions and potential enhancements.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.7
• /
• pp.1841-1857
• /
• 2023

CenterNet, a novel object detection algorithm without anchor based on key points, regards the object as a single center point for prediction and directly regresses the object's height and width. However, because the objects have different sizes, directly regressing their height and width will make the model difficult to converge and lose the intrinsic relationship between object's width and height, thereby reducing the stability of the model and the consistency of prediction accuracy. For this problem, we proposed an algorithm based on the regression of the diagonal half-length and the center angle, which significantly compresses the solution space of the regression components and enhances the intrinsic relationship between the decoded components. First, encode the object's width and height into the diagonal half-length and the center angle, where the center angle is the angle between the diagonal and the vertical centreline. Secondly, the predicted diagonal half-length and center angle are decoded into two length components. Finally, the position of the object bounding box can be accurately obtained by combining the corresponding center point coordinates. Experiments show that, when using CenterNet as the improved baseline and resnet50 as the Backbone, the improved model achieved 81.6% and 79.7% mAP on the VOC 2007 and 2012 test sets, respectively. When using Hourglass-104 as the Backbone, the improved model achieved 43.3% mAP on the COCO 2017 test sets. Compared with CenterNet, the improved model has a faster convergence rate and significantly improved the stability and prediction accuracy.

• Geomechanics and Engineering
• /
• v.11 no.2
• /
• pp.177-195
• /
• 2016

To secure the stability of geotechnical infrastructures and minimize failures during the construction process, a number of support systems have been introduced in the last several decades. In particular, stabilization methods using steel bars have been widely used in the field of geotechnical engineering. Rock bolt system is representative support system using steel bars. Pre-stressing has been applied to enhance reinforcement performance but can be released because of the failure of head or anchor sections. To overcome this deficiency, this paper proposes an innovative support system that can actively reinforce the weak ground along the whole structural element by introducing an active tension bolt containing a spring unit to the middle of the steel bar to increase its reinforcement capacity. In addition, the paper presents the support mechanism of the active tension bolt based on a theoretical study and employs an experimental study to validate the performance of the proposed active tension bolt based on a down-scaled model. To examine the feasibility of the active tension unit in a pillar, the paper considers a pullout test and a small-scale experimental model. The experimental results suggest the active tension bolt to be an effective support system for pillar reinforcement.

• Structural Engineering and Mechanics
• /
• v.63 no.1
• /
• pp.125-136
• /
• 2017

This study investigated the performance of a steel column standing on a reinforced concrete footing when it was subjected to collision of an eight-ton single unit truck. Finite element analyses of the structure with different connection schemes were performed using the finite element model of the truck, and the results showed that the behavior of the column subjected to the automobile impact depended largely on the column-footing connection detail. Various reinforcement schemes were investigated to mitigate the damage caused by the car impact. The probability of the model reinforced with a certain scheme to reach a given limit state was obtained by fragility analysis, and the effects of the combined reinforcement methods were investigated based on the equivalent fragility scheme. The analysis results showed that the reinforcement schemes such as increase of the pedestal area, decrease of the pedestal height, and the steel plate jacketing of the pedestal were effective in reducing the damage. As the speed of the automobile increased the contribution of the increase in the number of the anchor bolts and the dowel bars became more important to prevent crushing of the pedestal.

• Journal of Korean Society of Steel Construction
• /
• v.12 no.4 s.47
• /
• pp.445-455
• /
• 2000

This study proposed to beam-column joint model consisting of different type structural member to develop new structural system in the structural viewpoint as to a method to overcome various problem according to change of construction environment. This study promoted rigidity and capacity to stiffen reinforced concrete for steel structure end to increase rigidity of long spaned steel beam, and welt to steel flange to anchor U-shaped main bar of SRC structure end to easy stress flow between the different type structure. Through the series of experiments, proposed to possibility of this joint model, and investigated joint rigidity and capacity.

• Ocean Systems Engineering
• /
• v.9 no.1
• /
• pp.21-42
• /
• 2019

In this paper, the Finite-Analytic Navier-Stokes (FANS) code is coupled with an in-house finite-element code to study the dynamic interaction between a floating buoy and its mooring system. Hydrodynamic loads on the buoy are predicted with the FANS module, in which Large Eddy Simulation (LES) is used as the turbulence model. The mooring lines are modeled based on a slender body theory. Their dynamic responses are simulated with a nonlinear finite element module, MOORING3D. The two modules are coupled by transferring the forces and displacements of the buoy and its mooring system at their connections through an interface module. A free-decay model test was used to calibrate the coupled method. In addition, to investigate the capability of the present coupled method, numerical simulations of two degree-of-freedom vortex-induced motion of a CALM buoy in uniform currents were performed. With the study it can be verified that accurate predictions of the motion responses and tension responses of the CALM buoy system can be made with the coupling CFD-FEM method.

• Journal of Korea Multimedia Society
• /
• v.25 no.2
• /
• pp.335-344
• /
• 2022

In recent years, Internet Word of Mouth (IWOM) as an important way for internet consumers to obtain product information, it has become a key factor influencing consumers purchase intention. However, does IWOM also affect consumers' willingness to buy in the E-commerce live broadcast environment? What online IWOM can make a difference? And how? This study establishes a theoretical model of the impact of IWOM on consumer purchase intention in Chinese E-commerce live broadcast. Questionnaire data were collected from 204 people through the internet, with the model and the hypotheses are tested by SPSS 21.0 software. The research results show that: Firstly, the influencing factors of IWOM in the E-commerce live broadcast are different from those in general E-commerce. Secondly, anchor traits, word of mouth quantity, and product involvement have a positive impact on consumers purchase intentions, but the strength of relationship, word of mouth quality and Word of mouth timeliness do not have significantly direct effect on consumers' purchase intentions. Finally, conclusion with theoretical references and practical suggestions for E-commerce live broadcast merchants that may enter the Chinese market.