• Journal of Advanced Navigation Technology
• /
• v.16 no.6
• /
• pp.910-919
• /
• 2012

In this paper, we described development of on-board image processing system and its process and verified its performance through flight experiment. The image processing board has single ARM(Advanced Risk Machine) processor. We performed Embedded Linux Porting. Algorithm to be applied for object tracking is color-based image processing algorithm, it can be designed to track the object that has specific color on ground in real-time. To verify performance of the on-board image processing system, we performed flight test using the PNUAV, UAV developed by LAB. Also, we performed optimization of the image processing algorithm and kernel to improve real-time performance. Finally we confirmed that proposed system can track the blue-color object within four pixels error range consistently in the experiment.

• Journal of IKEEE
• /
• v.14 no.3
• /
• pp.224-231
• /
• 2010

Automatic detection of objects in images has been one of core challenges in the areas such as computer vision and pattern analysis. Especially, with the recent deployment of personal mobile devices such as smart phone, such technology is required to be transported to them. Usually, these smart phone users are equipped with devices such as camera, GPS, and gyroscope and provide various services through user-friendly interface. However, the smart phones fail to give excellent performance due to limited system resources. In this paper, we propose a new scheme to improve object recognition performance based on pre-computation and simple local features. In the pre-processing, we first find several representative parts from similar type objects and classify them. In addition, we extract features from each classified part and train them using regression functions. For a given query image, we first find candidate representative parts and compare them with trained information to recognize objects. Through experiments, we have shown that our proposed scheme can achieve resonable performance.

• Journal of the Korea Society for Simulation
• /
• v.31 no.2
• /
• pp.11-20
• /
• 2022

There is a critical need for AI assistance in guard operations of Army base perimeters, which is exacerbated by changes in the national defense and security environment such as force reduction. In addition, the possibility for human error inherent to perimeter guard operations attests to the need for an innovative revamp of current systems. The purpose of this study is to propose a real-time object detection AI tailored to military CCTV surveillance with three unique characteristics. First, training data suitable for situations in which relatively small objects must be recognized is used due to the characteristics of military CCTV. Second, we utilize a data augmentation algorithm suited for military context applied in the data preparation step. Third, a noise reduction algorithm is applied to account for military-specific situations, such as camouflaged targets and unfavorable weather conditions. The proposed system has been field-tested in a real-world setting, and its performance has been verified.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.1
• /
• pp.25-32
• /
• 2018

Due to heavy computational cost, deformable object simulation requires more effective collision detection method than rigid body simulation. However, when the CPU-based collision detection algorithm is purely applied to the GPU environment, the collision detection algorithm and the data structure optimized for the GPU environment are essential because the performance of the GPU can not be used properly. Therefore, we propose a GPU-based parallel collision detection algorithm for mass-spring system which is widely used for deformable object representation in this paper. The proposed method uses a parallel algorithm and data structure to reduce collision detection cost through GPU-based curling algorithm using AABB-Octree structure. In this paper, we prove the effectiveness of the proposed method by comparing the intersection test of all triangle pairs in parallel. The results of experimental tests show that the proposed method improves the performance by about 24% on average. Therefore, it is expected that the proposed method can improve the performance of real-time simulation for deformable objects.

• Korean Journal of Artificial Intelligence
• /
• v.12 no.4
• /
• pp.27-33
• /
• 2024

Various techniques are being researched to effectively detect forest fires. Among them, techniques using object detection models can monitor forest fires over wide areas 24 hours a day. However, detecting forest fires early with traditional object detection models is a very challenging task. While they show decent accuracy for thick smoke and large fires, they show low accuracy for faint smoke and small fires, and frequently generate false positives for lights that are like fires. In this paper, to solve these problems, we focus on leveraging local characteristics such as contours and textures of fire and smoke, which are crucial for accurate detection. Based on this approach, we propose EDAM (Edge driven Attention Module) that performs enhancement by richly utilizing contour and texture information of fire and smoke. EDAM extracts important edge information to generate feature maps with emphasized contour and texture information, and based on this map, performs Attention Mechanism to emphasize key characteristics of smoke and fire. Through this mechanism, the overall model performance was improved, with AP_sincreasing from 0.154 to 0.204 and AP_0.5 from 0.779 to 0.784, resulting in a significant improvement in AP_svalue to 32.47%. In practice, the model applying this technique showed excellent inference speed while greatly improving detection performance for small objects compared to existing models and reduced false positive rates for building and street light illumination in nighttime environments that are easily mistaken for fire.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.38 no.4
• /
• pp.353-361
• /
• 2020

Due to the rapid urbanization, various traffic problems such as traffic jams during commute and regular traffic jams are occurring. In order to solve these traffic problems, it is necessary to quickly and accurately estimate and analyze traffic volume. ITS (Intelligent Transportation System) is a system that performs optimal traffic management by utilizing the latest ICT (Information and Communications Technology) technologies, and research has been conducted to analyze fast and accurate traffic volume through various techniques. In this study, we proposed a deep learning-based vehicle detection method using UAV (Unmanned Aerial Vehicle) video for real-time traffic analysis with high accuracy. The UAV was used to photograph orthogonal videos necessary for training and verification at intersections where various vehicles pass and trained vehicles by classifying them into sedan, truck, and bus. The experiment on UAV dataset was carried out using YOLOv3 (You Only Look Once V3), a deep learning-based object detection technique, and the experiments achieved the overall object detection rate of 90.21%, precision of 95.10% and the recall of 85.79%.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.4 no.2
• /
• pp.108-113
• /
• 2003

Research results have demonstrated that conventional client-server databases have scalability problem in the presence of many concurrent clients. The multi-tier architecture that exploits similarities in clients' object access behavior partitions clients into logical clusters according to their object request pattern. As a result, object requests that are served inside the clusters, server load and request response time can be optimized. Management of clustering by utilizing clients' access pattern-based is an important component for the system's goal. Off-line methods optimizes the quality of the global clustering, the necessary cost and clustering schedule needs to be considered and planned carefully in respect of stable system's performance. In this paper, we propose methods that detect changes in access behavior and optimize system configuration in real time. Finally this paper demonstrates the effectiveness of on-line change detection and results of experimental investigation concerning reconfiguration.

• Journal of IKEEE
• /
• v.17 no.3
• /
• pp.371-377
• /
• 2013

This thesis propose an efficient parallel structure method for implementing a FAST and BRIEF algorithm based Augmented Reality. SURF algorithm that is well known in the object recognition algorithms is robust in object recognition. However, there is a disadvantage for real time operation because, SURF implementation requires a lot of computation. Therefore, we used a FAST and BRIEF algorithm for object recognition, and we improved Conventional Parallel Structure based on OpenMP Library. As a result, it achieves a 70%~100% improvement in execution time on the embedded system.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.5
• /
• pp.237-242
• /
• 2023

Research on high-performance walking robots is being actively conducted, and quadruped walking robots are receiving a lot of attention due to their excellent mobility and adaptability on uneven terrain, but they are difficult to introduce and utilize due to high cost. In this paper, to increase utilization by applying intelligent functions to a low-cost quadruped robot, we present a method of improving uneven terrain overcoming ability by mounting IMU and reinforcement learning on embedded board and automatically detecting objects using camera and deep learning. The robot consists of the legs of a quadruped mammal, and each leg has three degrees of freedom. We train complex terrain in simulation environments with designed 3D model and apply it to real robot. Through the application of this research method, it was confirmed that there was no significant difference in walking ability between flat and non-flat terrain, and the behavior of performing person detection in real time under limited experimental conditions was confirmed.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2007.10a
• /
• pp.293-297
• /
• 2007

This thesis suggests a new algorithm to detects multiple moving objects using a CMODE(Correct Multiple Object DEtection) method in the color images acquired in real-time and to track the interested pedestrian using motion and hue information. The multiple objects are detected, and then shaking trees or moving cars are removed using structural characteristics and shape information of the man , the interested pedestrian can be detected, The first similarity judgment for tracking an interested pedestrian is to use the distance between the previous interested pedestrian's centroid and the present pedestrian's centroid. For the area where the first similarity is detected, three feature points are calculated using k-mean algorithm, and the second similarity is judged and tracked using the average hue value for the $3{\times}3$ area of each feature point. The zooming of camera is adjusted to track an interested pedestrian at a long distance easily and the FOV(Field of View) of camera is adjusted in case the pedestrian is not situated in the fixed range of the screen. As a experiment results, comparing the suggested CMODE method with the labeling method, an average approach rate is one fourth of labeling method, and an average detecting time is faster three times than labeling method. Even in a complex background, such as the areas where trees are shaking or cars are moving, or the area of shadows, interested pedestrian detection is showed a high detection rate of average 96.5%. The tracking of an interested pedestrian is showed high tracking rate of average 95% using the information of situation and hue, and interested pedestrian can be tracked successively through a camera FOV and zooming adjustment.