Chang-sung Jeong

Dr. Chang-sung Jeong


Research Focus & Metrics

Academic Career Length

37Years (1990-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 June 2018 in EURASIP Journal on Image and Video Processing
Verified
Journal Article
Open access
PDF file
YKYoon-Ki Kim
Yoon-Ki Kim
YKYongsung Kim
Yongsung Kim
CJChang-sung Jeong
Chang-sung Jeong
As the demand for real-time data processing increases, a high-speed processing platform for large-scale stream data becomes necessary. For fast processing large-scale stream data, it is essential to use multiple distributed nodes. So far, there have been few studies on real-time massive image processing through efficient management and allocation of heterogeneous resources for various user-specified nodes on distributed environments. In this paper, we shall present a new platform called RIDE (Real-time massive Image processing platform on Distributed Environment) which efficiently allocates resources and executes load balancing according to the amount of stream data on distributed environments. It minimizes communication overhead by using a parallel processing strategy which handles the stream data considering both coarse-grained and fine-grained parallelism simultaneously. Coarse-grained parallelism is achieved by the automatic allocation of input streams onto partitions of broker buffer each processed by its corresponding worker node, and maximized by adaptive resource management which adjusts the number of worker nodes in a group according to the frame rate in real time. Fine-grained parallelism is achieved by parallel processing of task on each worker node and maximized by allocating heterogeneous resources such as GPU and embedded machines appropriately. Moreover, it provides a scheme of application topology which has a great advantage for higher performance by configuring the worker nodes of each stage using adaptive heterogeneous resource management. Finally, it supports dynamic fault tolerance for real-time image processing through the coordination between components in our system.

1 October 2010
Verified
Conference Paper
Closed access
IJIn-Yong Jung
In-Yong Jung
CJChang-sung Jeong
Chang-sung Jeong
This paper proposes a new connected component labeling algorithm for GPGPU applications based on NVIDIA's CUDA. Various approaches and algorithms for connected component labeling with minimal execution time were designed, but the most of them have been focused on optimizing CPU algorithm. Therefore it is hard to apply these approaches to GPGPU programming models such as NVIDIA's CUDA. Today, GPGPU (General Purpose Graphic Processing Unit) technologies offer dedicated parallel hardware and programming model, and many applications are being moved onto the GPGPU. This algorithm is a multi-pass algorithm to utilize for GPGPU applications, and evaluation results show that maximum speedup is more than double compared with conventional CPU algorithms.

6 January 2015 in Cluster Computing
Verified
Journal Article
Closed access
PDF file
DSDongBum Seo
DongBum Seo
CJChang-sung Jeong
Chang-sung Jeong
YJYou-Boo Jeon
You-Boo Jeon
KLKeun-Ho Lee
Keun-Ho Lee

6 March 2013 in EURASIP Journal on Wireless Communications and Networking
Verified
Journal Article
Open access
PDF file
BHByong John Han
Byong John Han
IJIn-Yong Jung
In-Yong Jung
KKKi-Hyun Kim
Ki-Hyun Kim
DLDo-Kwang Lee
Do-Kwang Lee
SRSeungmin Rho
Seungmin Rho
CJChang-sung Jeong
Chang-sung Jeong
In this article, a new content-centric collaboration platform called active content collaboration platform which supports automated content-centric collaboration on cloud system is presented. It supports event-driven automatic collaboration by specifying each active work based on active work description language and automating the execution of collaborative task flow composed of active works. It supports a modularized and extensible architecture by making its components in separate modules. Also, it provides a scalable high-performance architecture by supporting multi-level active work processing in active work execution engine, and allocating virtual machines for the computation intensive high-level action such as volume rendering through auto-scalable allocation on cloud system. For the experiment and evaluation, we shall show the results of implementing the collaborative medical application on our system where computation intensive application like a volume rendering is used for MRI analysis.

7 March 2023 in Sensors
Verified
Journal Article
Open access
PDF file
BKBoSeon Kang
BoSeon Kang
CJChang-sung Jeong
Chang-sung Jeong
Due to the sharp increase in household waste, its separate collection is essential in order to reduce the huge amount of household waste, since it is difficult to recycle trash without separate collection. However, since it is costly and time-consuming to separate trash manually, it is crucial to develop an automatic system for separate collection using deep learning and computer vision. In this paper, we propose two Anchor-free-based Recyclable Trash Detection Networks (ARTD-Net) which can recognize overlapped multiple wastes of different types efficiently by using edgeless modules: ARTD-Net1 and ARTD-Net2. The former is an anchor-free based one-stage deep learning model which consists of three modules: centralized feature extraction, multiscale feature extraction and prediction. The centralized feature extraction module in backbone architecture focuses on extracting features around the center of the input image to improve detection accuracy. The multiscale feature extraction module provides feature maps of different scales through bottom-up and top-down pathways. The prediction module improves classification accuracy of multiple objects based on edge weights adjustments for each instance. The latter is an anchor-free based multi-stage deep learning model which can efficiently finds each of waste regions by additionally exploiting region proposal network and RoIAlign. It sequentially performs classification and regression to improve accuracy. Therefore, ARTD-Net2 is more accurate than ARTD-Net1, while ARTD-Net1 is faster than ARTD-Net2. We shall show that our proposed ARTD-Net1 and ARTD-Net2 methods achieve competitive performance in mean average precision and F1 score compared to other deep learning models. The existing datasets have several problems that do not deal with the important class of wastes produced commonly in the real world, and they also do not consider the complex arrangement of multiple wastes with different types. Moreover, most of the existing datasets have an insufficient number of images with low resolution. We shall present a new recyclables dataset which is composed of a large number of high-resolution waste images with additional essential classes. We shall show that waste detection performance is improved by providing various images with the complex arrangement of overlapped multiple wastes with different types.

2 June 2022 in Symmetry
Verified
Journal Article
Open access
PDF file
WLWon-Chan Lee
Won-Chan Lee
YJYou-Boo Jeon
You-Boo Jeon
SHSeong-Soo Han
Seong-Soo Han
CJChang-sung Jeong
Chang-sung Jeong
This paper deals with the prediction of the future location of vehicles, which is attracting attention in the era of the fourth industrial revolution and is required in various fields, such as autonomous vehicles and smart city traffic management systems. Currently, vehicle traffic prediction models and accident prediction models are being tested in various places, and considerable progress is being made. However, there are always errors in positioning when using wireless sensors due to various variables, such as the appearance of various substances (water, metal) that occur in the space where radio waves exist. There have been various attempts to reduce the positioning error in such an Internet of Things environment, but there is no definitive method with confirmed performance. Of course, location prediction is also not accurate. In particular, since a vehicle moves rapidly in space, it is increasingly affected by changes in the environment. Firstly, it was necessary to develop a spatial positioning algorithm that can improve the positioning accuracy. Secondly, for the data generated by the positioning algorithm, a machine learning method suitable for position prediction was developed. Based on the above two developed algorithms, through experiments, we found a means to reduce the error of positioning through radio waves and to increase the accuracy of positioning. We started with the idea of changing the positioning space itself from a three-dimensional space into a two-dimensional one. With changes in the time and space of radio wave measurement, the location was measured by transforming the spatial dimension to cope with environmental changes. This is a technology that predicts a location through machine learning on time series data using a direction angle classification technique. An experiment was conducted to verify the performance of the proposed technology. As a result, the accuracy of positioning was improved, and the accuracy of location prediction increased in proportion to the learning time. It was possible to confirm the prediction accuracy increase of up to 80% with changes. Considering that the accuracy result for location prediction presented by other researchers is 70%, through this study, the result was improved by 10% compared to the existing vehicle location prediction accuracy. In conclusion, this paper presents a positioning algorithm and machine learning methodology for vehicle positioning. By proving its usefulness through experiments, this study provides other researchers with a new definition of space for predicting the location of a vehicle, and a machine learning method using direction angles.