• ETRI Journal
• /
• v.46 no.2
• /
• pp.307-322
• /
• 2024

Most object detection methods use a horizontal bounding box that causes problems between adjacent objects with arbitrary directions, resulting in misaligned detection. Hence, the horizontal anchor should be replaced by a rotating anchor to determine oriented bounding boxes. A two-stage process of delineating a horizontal bounding box and then converting it into an oriented bounding box is inefficient. To improve detection, a box-boundary-aware vector can be estimated based on a convolutional neural network. Specifically, we propose a ResNeXt101 encoder to overcome the weaknesses of the conventional ResNet, which is less effective as the network depth and complexity increase. Owing to the cardinality of using a homogeneous design and multi-branch architecture with few hyperparameters, ResNeXt captures better information than ResNet. Experimental results demonstrate more accurate and faster oriented object detection of our proposal compared with a baseline, achieving a mean average precision of 89.41% and inference rate of 23.67 fps.

• Journal of Institute of Convergence Technology
• /
• v.6 no.1
• /
• pp.23-26
• /
• 2016

DC motor shafts have several defects such as double cut, deep scratch on surface, and defects in diameter and length. The deep scratches are due to collision among the other shafts. So the scratches are long and thin but their orientations are random. If the smallest enclosing box, i.e. oriented bounding box for a detective point group is found, then the size of the corresponding defect can be modeled as its diagonal length. This paper proposes an suface inspection algorithm for the DC motor shaft using the oriented bounding box. To evaluate the proposed algorithm, a test bed is made with a line scan CCD camera (4096 pixels/line) and two rollers mechanism to rotate the shaft. The experimental result on a pre-processed image with contrast streching algorithm, shows that the proposed algorithm sucessfully finds 150 surface defects and its computation time (0.291 msec) is enough fast for the requirement (4 seconds).

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.2402-2406
• /
• 2005

A collision query determines the intersection between given objects, and is used in computer-aided design and manufacturing, animation and simulation systems, and physically-based modeling. Bounding volume hierarchies are one of the simplest and most widely used data structures for performing collision detection on complex models. In this paper, we present hierarchy of oriented rounded bounding volume for fast proximity queries. Designing hierarchies of new bounding volumes, we use to combine multiple bounding volume types in a single hierarchy. The new bounding volume corresponds to geometric shape composed of a core primitive shape grown outward by some offset such as the Minkowski sum of rectangular box and a sphere shape. In the experiment of parallel close proximity, a number of benchmarks to measure the performance of the new bounding box and compare to that of other bounding volumes.

• Journal of the Korea Computer Graphics Society
• /
• v.13 no.4
• /
• pp.21-27
• /
• 2007

The 3D Delaunay triangulation is the most widely used method for the mesh creation via the triangulation of a 3D point cloud. However, the method involves a heavy computational cost and, hence, in many interactive applications, it is not appropriate for surface triangulation. In this paper, we propose an efficient triangulation method to create a surface mesh from a 3D point cloud. We divide a set of object points into multiple subsets and apply the 2D Delaunay triangulation to each subset. A given 3D point cloud is cut into slices with respect to the OBB(Oriented Bounding Box) of the point set. The 2D Delaunay triangulation is applied to each subset producing a partial triangulation. The sum of the partial triangulations constitutes the global mesh. As a postprocessing process, we eliminate false edges introduced in the split steps of the triangulation and improve the results. The proposed method can be effectively applied to various image-based modeling applications.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2011.11a
• /
• pp.459-460
• /
• 2011

본 논문에서는 이러한 방식에서 탈피한, 3D 그래픽을 이용한 1인칭 미니 회피게임을 제작하는 방법에 대해 제안한다. 기존의 3인칭 시점에 상하좌우 네 방향으로만 움직이는 방식을 벗어나, 시점을 1인칭으로 변환하고 FPS와 같은 시점과 이동방식을 제공하며, 기존의 2D게임에서 사용되던 축이 고정된 오브젝트의 충돌인 AABB(Axis Aligned Bounding Box)가 아닌 축이 수시로 변하는 OBB(Oriented Bounding Box) 방식을 사용함으로 써, 3D 그래픽에서도 2D 그래픽에서처럼 정교한 충돌 검출 기능이 가능하도록 제작한다.

• Journal of KIISE:Computer Systems and Theory
• /
• v.35 no.1
• /
• pp.44-52
• /
• 2008

In this paper we consider detecting collisions between characters whose motion is specified by motion capture data. Since we are targeting on massive crowd simulation, we only consider rough collisions, modeling the characters as a disk in the floor plane. To provide efficient collision detection, we introduce a hierarchical bounding volume, the Motion Oriented Bounding Box tree (MOBB tree). A MOBBtree stores space-time bounds of a motion clip. In crowd animation tests, MOBB trees performance improvements ranging between two and an order of magnitude.

• Journal of the Optical Society of Korea
• /
• v.15 no.2
• /
• pp.152-154
• /
• 2011

We propose a novel method which detects convex and concave corners using radial projection. The sum of two neighbors' differences at the local maxima or minima of the radial projection is compared with the angle threshold for detecting corners. In addition, the use of oriented bounding box trees and partial circles makes it possible to detect the corners of complex shapes. The experimental results show that the proposed method can separately detect the convex and concave corners, and that this method is scale invariant.

• International Journal of Computer Science & Network Security
• /
• v.24 no.5
• /
• pp.129-134
• /
• 2024

Object Detection is an emerging technology in the field of Computer Vision and Image Processing that deals with detecting objects of a particular class in digital images. It has considered being one of the complicated and challenging tasks in computer vision. Earlier several machine learning-based approaches like SIFT (Scale-invariant feature transform) and HOG (Histogram of oriented gradients) are widely used to classify objects in an image. These approaches use the Support vector machine for classification. The biggest challenges with these approaches are that they are computationally intensive for use in real-time applications, and these methods do not work well with massive datasets. To overcome these challenges, we implemented a Deep Learning based approach Convolutional Neural Network (CNN) in this paper. The Proposed approach provides accurate results in detecting objects in an image by the area of object highlighted in a Bounding Box along with its accuracy.

• The KIPS Transactions:PartB
• /
• v.15B no.1
• /
• pp.25-36
• /
• 2008

In this paper, we present a robust hand recognition approach to sudden illumination changes. The proposed approach constructs a background model with respect to hue and hue gradient in HSI color space and extracts a foreground hand region from an input image using the background subtraction method. Eighteen features are defined for a hand pose and multi-class SVM(Support Vector Machine) approach is applied to learn and classify hand poses based on eighteen features. The proposed approach robustly extracts the contour of a hand with variations in illumination by applying the hue gradient into the background subtraction. A hand pose is defined by two Eigen values which are normalized by the size of OBB(Object-Oriented Bounding Box), and sixteen feature values which represent the number of hand contour points included in each subrange of OBB. We compared the RGB-based background subtraction, hue-based background subtraction and the proposed approach with sudden illumination changes and proved the robustness of the proposed approach. In the experiment, we built a hand pose training model from 2,700 sample hand images of six subjects which represent nine numerical numbers from one to nine. Our implementation result shows 92.6% of successful recognition rate for 1,620 hand images with various lighting condition using the training model.