• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.05a
• /
• pp.641-642
• /
• 2022

단일 데이터로부터의 이동 객체에 대한 행동 인식 연구는 데이터 수집 과정에서 발생하는 노이즈의 영향을 크게 받는다. 본 논문은 영상 데이터와 센서 데이터를 이용하여 다중 융합 네트워크 기반 이동 객체 행동 인식 방법을 제안한다. 영상으로부터 객체가 감지된 영역의 추출과 센서 데이터의 이상치 제거 및 결측치 보간을 통해 전처리된 데이터들을 융합하여 시퀀스를 생성한다. 생성된 시퀀스는 CNN(Convolutional Neural Networks)과 LSTM(Long Short Term Memory)기반 다중 융합 네트워크 모델을 통해 시계열에 따른 행동 특징들을 추출하고, 깊은 FC(Fully Connected) 계층을 통해 특징들을 융합하여 행동을 예측한다. 본 연구에서 제시된 방법은 사람을 포함한 동물, 로봇 등의 다양한 객체에 적용될 수 있다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.175-178
• /
• 2023

본 논문은 영화의 스토리 인식을 위해 인물 간 상호작용 중 물리적 상호작용 즉, 물리접촉을 검출하는 방법을 제안한다. YOLO를 사용해 영상에서 인간객체를 탐지하고, Mediapipe를 사용해 골격 감지를 진행함으로써 인물의 뼈대를 랜드마크화 하고 타 객체 간의 랜드마크가 일정값 이하로 내려오면 Threshold를 적용해 객체 간의 물리적 접촉을 판단한다, 실험 결과, 50개 17,741 frame의 영상에서 정확도 99.66%의 정밀도 77.27%, 재현율 62.38%로 모델의 전반적인 성능을 나타내는 F1점수는 69%로 나타났다.

• Journal of Internet Computing and Services
• /
• v.23 no.5
• /
• pp.47-58
• /
• 2022

In recent years, navigation aids, which plays as sea traffic lights, have been digitized, and are developing beyond simple sign purpose to provide various functions such as marine information collection, supervision, control, etc. For example, Busan Port which is located in South Korea is leading the application of the advanced technologies by installing cameras on buoys and recording video images to supervise maritime accidents. However, there are difficulties to perform their major functions since the advanced technologies require long-term battery operation and also management and maintenance of them are hampered by marine characteristics. This study proposes a system that can automatically notify maritime objects passing around buoys by analyzing image information. In the existing sensor-based accident prevention systems, the alarms are generated by a collision detection sensor. The system can identify the cause of the accident whilst even though it is difficult not possible to fundamentally prevent the accidents. Therefore, in order to overcome these limitations, the proposed a maritime object detection system is based on marine characteristics. The experiments demonstrate that the proposed system shows about 5 times faster processing speed than other existing algorithms.

• Transactions on Semiconductor Engineering
• /
• v.2 no.1
• /
• pp.17-20
• /
• 2024

This paper presents a novel lightweight object detection model tailored for low-powered edge devices, addressing the limitations of traditional resource-intensive computer vision models. Our proposed detector, inspired by the Single Shot Detector (SSD), employs a compact yet robust network design. Crucially, it integrates an 'enhancer block' that significantly boosts its efficiency in detecting smaller objects. The model comprises two primary components: the Light_Block for efficient feature extraction using Depth-wise and Pointwise Convolution layers, and the Enhancer_Block for enhanced detection of tiny objects. Trained from scratch on the Udacity Annotated Dataset with image dimensions of 300x480, our model eschews the need for pre-trained classification weights. Weighing only 5.5MB with approximately 0.43M parameters, our detector achieved a mean average precision (mAP) of 27.7% and processed at 140 FPS, outperforming conventional models in both precision and efficiency. This research underscores the potential of lightweight designs in advancing object detection for edge devices without compromising accuracy.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.1098-1099
• /
• 2023

다중객체 분석 모델 Yolo를 기반으로 물체를 감지하기 위해서 학습을 진행하고 학습을 통해서 얻어낸 모델을 기반으로 드론을 통해서 얻어낸 영상을 통해 컨테이너 파손이 된 부분을 감지하는 프로젝트를 진행했다.

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.8
• /
• pp.2233-2244
• /
• 1999

It is essential to guarantee high reliability of embedded real-time systems since the failure of such systems may result in large financial damage or threaten human life. Though many researches have devoted to fault tolerant mechanisms, most of them are object-level fault tolerant mechanisms that can detect errors occurred in a single object and treat the errors in object-level. As embedded real-time systems become more complex and larger, there exist faults that cannot be detected by or tolerated with object-level fault tolerance. Hence, system-level fault tolerance is needed. System-level fault tolerance examines the status of a system whether the system is normal or not by analyzing the status of objects. When an error is detected it should be capable of locating the fault and performing an appropriate recovery and reconfiguration action. In this paper, we propose RobustRTO(Robust Real-Time Object) that provides object-level fault tolerance capability and RMO(Region Monitor real-time Object) that offers system-level fault tolerance capability. Then we show how highly dependable fault tolerant systems can be modeled by RobustRTO and RMO. The model is presented based on real-time objects.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.39B no.5
• /
• pp.271-280
• /
• 2014

In wireless sensor networks, there have been two methods for sensing continuously moving object tracking: user-query based method and periodic report based method. Although the former method requires overhead for user query rather than the latter method, the former one is known as an energy-efficient method without transferring unnecessary information. In the former method, a virtual tree, consisting of sensor nodes, is exploited for the user querying and sensor reporting. The tree stores the information about mobile objects; the stored information is triggered to report by the user query. However, in case of fast moving object, the tracking accuracy reduces due to the time delay of end-to-end repeated query. To solve the problem, we propose a query relaying method reducing the time delay for mobile object tracking. In the proposed method, the nodes in the tree relay the query to the adjacent node according to the movement of mobile object tracking. Relaying the query message reduces the end-to-end querying time delay. Simulation results show that our method is superior to the existing ones in terms of tracking accuracy.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.10 no.5
• /
• pp.229-235
• /
• 2010

In this paper, we propose a method to improve the efficiency of the moving object detection in real-time surveillance camera system. The existing methods, the methods using differential image and background image, are difficult to detect the moving object from outside the video streams. The proposed method keeps the background image if it doesn't be detected moving object using the differential value between a previous frame and a current frame. And the background image is renewed as the moving object is gone in a frame. To decide people and wild pig, the proposed system estimates a bounding box enclosing each moving object in the detecting region. As a result of simulation, the proposed method is better than the existing method.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.69-71
• /
• 2022

Today, the AI market is very large due to the development of AI. Among them, the most advanced AI is image detection. Thus, there are many object detection models using YOLOv5.However, most object detection in AI is focused on detecting objects that are stereotyped.In order to recognize such unstructured data, the object may be recognized by learning and filtering the object. Therefore, in this paper, a fire monitoring system using YOLOv5 was designed to detect and analyze unstructured data fires and suggest ways to improve the fire object detection model.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.11
• /
• pp.789-794
• /
• 2013

A vehicle tracking system makes it possible to induce the vehicle movement path for avoiding traffic congestion and to prevent traffic accidents in advance by recognizing traffic flow, monitoring vehicles, and detecting road accidents. To track the vehicles effectively, those which appear in a sequence of video frames need to identified by extracting the features of each object in the frames. Next, the identical vehicles over the continuous frames need to be recognized through the matching among the objects' feature values. In this paper, we identify objects by binarizing the difference image between a target and a referential image, and the labelling technique. As feature values, we use the center coordinate of the minimum bounding rectangle(MBR) of the identified object and the averages of 1D FFT(fast Fourier transform) coefficients with respect to the horizontal and vertical direction of the MBR. A vehicle is tracked in such a way that the pair of objects that have the highest similarity among objects in two continuous images are regarded as an identical object. The experimental result shows that the proposed method outperforms the existing methods that use geometrical features in tracking accuracy.