• 한국멀티미디어학회논문지
• /
• 제9권5호
• /
• pp.529-541
• /
• 2006

본 논문에서는 객체기반 비디오 부호화 또는 멀티미디어 편집을 위한 반지동 비디오 객체 분할방식을 제안한다. 반자동 객체분할은 사용자 지원에 의한 분할 방식으로, 비디오 시퀀스의 초기 프레임에서 사용자가 관심객체의 경계를 표시하고 이후의 영상 프레임의 객체를 배경으로부터 연속적으로 분리해 낸다. 제안된 방식은 부분적으로 사용자 조력에 의한 프레임내 분할과 완전 자동에 의한 프레임간 분할 처리과정으로 구성되는데, 영상 전체에 대해 연산을 수행하는 기존 방식과는 달리 객체 경계가 존재하는 영상영역 부분에서만 연산을 수행한다. 프레임내 분할은 사용자가 관심객체의 경계를 지정하고, 이 경계 주위 화소들의 유사성을 이용한 후처리에 의해 정확한 초기 객체를 구한다. 프레임간 분할에서는 이전 프레임에서 추출한 객체의 경계 정보에 근거하여 시간적 유사성을 구한 후 경계와 영역 추적에 의해 연속적으로 동영상 객체를 추출한다. 실험결과로부터 제안된 방식은 비디오 편집, 객체기반 비디오 압축 및 인덱싱 등의 멀미디어 응용에 사용 가능할 정도로 안정되고 정확한 객체추출을 수행함을 확인하였다. 이 결과를 바탕으로 다수의 편리한 기능을 포함한 비디오 편집시스템을 개발하였다.

• 대한임베디드공학회논문지
• /
• 제17권5호
• /
• pp.289-296
• /
• 2022

In this paper, we developed a deep learning-based recyclable object detection model. The model is developed based on YOLOv5 that is a one-stage detector. The deep learning model detects and classifies the recyclable object into 7 categories: paper, carton, can, glass, pet, plastic, and vinyl. We propose two methods for recyclable object detection models to solve problems during training. Bounding Box CutMix solved the no-objects training images problem of Mosaic, a data augmentation used in YOLOv5. Standardized Distance-based IoU replaced DIoU using a normalization factor that is not affected by the center point distance of the bounding boxes. The recyclable object detection model showed a final mAP performance of 0.91978 with Bounding Box CutMix and 0.91149 with Standardized Distance-based IoU.

• Journal of Astronomy and Space Sciences
• /
• 제34권1호
• /
• pp.31-35
• /
• 2017

Optical observation is one of the most common techniques used for characterizing the physical properties of unknown objects and debris in space. This research presents measurements and properties of the new object 96019 from ground-based optical methods. Optical observations of this small object were performed using a charge-coupled device (CCD) camera and the Santel-500 telescope at the Zvenigorod Observatory. The orbital elements and physical properties of this object, such as area-to-mass ratio, have been determined. The results show that this small object has a low area-to-mass ratio, between 0.009 and $0.12m^2/kg$. The light curve of object 96019 is given: Over the time intervals, variations in brightness are analyzed and the maximum brightness was found to be 12.4 magnitudes. The observational results show that, this object brightens by about three magnitudes over a time span of three minutes. Based on these observations, the characteristics and physical properties of this object are discussed.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송공학회 2009년도 IWAIT
• /
• pp.223-228
• /
• 2009

We propose an on-line machine learning approach for object recognition, where new images are continuously added and the recognition decision is made without delay. Random forest (RF) classifier has been extensively used as a generative model for classification and regression applications. We extend this technique for the task of building incremental component-based detector. First we employ object descriptor model based on bag of covariance matrices, to represent an object region then run our on-line RF learner to select object descriptors and to learn an object classifier. Experiments of the object recognition are provided to verify the effectiveness of the proposed approach. Results demonstrate that the propose model yields in object recognition performance comparable to the benchmark standard RF, AdaBoost, and SVM classifiers.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권10호
• /
• pp.4976-4994
• /
• 2018

The diffusion-based salient object detection methods have shown excellent detection results and more efficient computation in recent years. However, the current diffusion-based salient object detection methods still have disadvantage of detecting the object appearing at the image boundaries and different scales. To address the above mentioned issues, this paper proposes a multi-scale diffusion-based salient object detection algorithm with background and objectness seeds. In specific, the image is firstly over-segmented at several scales. Secondly, the background and objectness saliency of each superpixel is then calculated and fused in each scale. Thirdly, manifold ranking method is chosen to propagate the Bayessian fusion of background and objectness saliency to the whole image. Finally, the pixel-level saliency map is constructed by weighted summation of saliency values under different scales. We evaluate our salient object detection algorithm with other 24 state-of-the-art methods on four public benchmark datasets, i.e., ASD, SED1, SED2 and SOD. The results show that the proposed method performs favorably against 24 state-of-the-art salient object detection approaches in term of popular measures of PR curve and F-measure. And the visual comparison results also show that our method highlights the salient objects more effectively.

• 제어로봇시스템학회논문지
• /
• 제12권8호
• /
• pp.746-751
• /
• 2006

In this paper, we present tracking algorithm for the indoor moving object. We research passive method using a camera and image processing. It had been researched to use dynamic based estimators, such as Kalman Filter, Extended Kalman Filter and Particle Filter for tracking moving object. These algorithm have a good performance on real-time tracking, but they have a limit. If the shape of object is changed or object is located on complex background, they will fail to track them. This problem will need the complicated image processing algorithm. Finally, a large algorithm is made from integration of dynamic based estimator and image processing algorithm. For eliminating this inefficiency problem, image based estimator, Mean-shift Algorithm is suggested. This algorithm is implemented by color histogram. In other words, it decide coordinate of object's center from using probability density of histogram in image. Although shape is changed, this is not disturbed by complex background and can track object. This paper shows the results in real camera system, and decides 3D coordinate using the data from mean-shift algorithm and relationship of real frame and camera frame.

• 대한원격탐사학회지
• /
• 제23권3호
• /
• pp.153-160
• /
• 2007

Forest cover density studies using high resolution satellite data and object oriented classification are limited in India. This article focuses on the potential use of QuickBird satellite data and object oriented classification in forest density mapping. In this study, the high-resolution satellite data was classified based on NDVI/pixel based and object oriented classification methods and results were compared. The QuickBird satellite data was found to be suitable in forest density mapping. Object oriented classification was superior than the NDVI/pixel based classification. The Object oriented classification method classified all the density classes of forest (dense, open, degraded and bare soil) with higher producer and user accuracies and with more kappa statistics value compared to pixel based method. The overall classification accuracy and Kappa statistics values of the object oriented classification were 83.33% and 0.77 respectively, which were higher than the pixel based classification (68%, 0.56 respectively). According to the Z statistics, the results of these two classifications were significantly different at 95% confidence level.

• 한국멀티미디어학회논문지
• /
• 제20권6호
• /
• pp.893-910
• /
• 2017

In this paper, we propose a simple online multiple object (human) tracking method, LKDeep (Lucas-Kanade feature and Detection based Simple Online Multiple Object Tracker), which can run in fast online enough on CPU core only with acceptable tracking performance for embedded surveillance purpose. The proposed LKDeep is a pragmatic hybrid approach which tracks multiple objects (humans) mainly based on LK features but is compensated by detection on periodic times or on necessity times. Compared to other state-of-the-art multiple object tracking methods based on 'Tracking-By-Detection (TBD)' approach, the proposed LKDeep is faster since it does not have to detect object on every frame and it utilizes simple association rule, but it shows a good object tracking performance. Through experiments in comparison with other multiple object tracking (MOT) methods using the public DPM detector among online state-of-the-art MOT methods reported in MOT challenge [1], it is shown that the proposed simple online MOT method, LKDeep runs faster but with good tracking performance for surveillance purpose. It is further observed through single object tracking (SOT) visual tracker benchmark experiment [2] that LKDeep with an optimized deep learning detector can run in online fast with comparable tracking performance to other state-of-the-art SOT methods.

• 지능정보연구
• /
• 제9권3호
• /
• pp.99-115
• /
• 2003

본 논문은 지식 기반 객체지 향 공간 데이터베이스시스템 KOBOS를 제안한다. 객체지향 공간 데이터베이스 시스템의 데이터 모델링과 근접 질의답변에 지식기반 접근법을 도입한다. 공간객체와 근접 공간 연산자를 다루기 위해 다음과 같은 세 단계 객체지향 데이터 모델을 제안하고 있다: (1) 공간 형상 모델; (2) 공간 객체 모델: (3) 내부 기술 모델. 근접 공간 연산자의 범위는 공간 타입 추상 계층으로 알 수 있다. 또한 객체지향 공간 질의어인 SOQL을 제안한다. SOQL은 공간 객체의 다양한 출력과 공간 및 비 공간 객체의 검색을 수행할 수 있는 통합 기능을 제공해준다. 효율적인 혼합 질의 처리를 위하여, 하향 공간 질의 처리 방법을 이용하여 처리해 준다.

• 로봇학회논문지
• /
• 제17권3호
• /
• pp.255-263
• /
• 2022

3D object detection pipelines often incorporate RGB-based object detection methods such as YOLO, which detects the object classes and bounding boxes from the RGB image. However, in complex environments where objects are heavily cluttered, bounding box approaches may show degraded performance due to the overlapping bounding boxes. Mask based methods such as Mask R-CNN can handle such situation better thanks to their detailed object masks, but they require much longer time for data preparation compared to bounding box-based approaches. In this paper, we present a 3D object recognition pipeline which uses either the YOLO or Mask R-CNN real-time object detection algorithm, K-nearest clustering algorithm, mask reduction algorithm and finally Principal Component Analysis (PCA) alg orithm to efficiently detect 3D poses of objects in a complex environment. Furthermore, we also present an improved YOLO based 3D object detection algorithm that uses a prioritized heightmap clustering algorithm to handle overlapping bounding boxes. The suggested algorithms have successfully been used at the Artificial-Intelligence Robot Challenge (ARC) 2021 competition with excellent results.