• IEMEK Journal of Embedded Systems and Applications
• /
• v.12 no.2
• /
• pp.113-120
• /
• 2017

In this paper, we present a night-time vehicle detection method using CNN (Convolutional Neural Network) classification. The camera based night-time vehicle detection plays an important role on various advanced driver assistance systems (ADAS) such as automatic head-lamp control system. The method consists mainly of thresholding, labeling and classification steps. The classification step is implemented by existing CIFAR-10 model CNN. Through the simulations tested on real road video, we show that CNN classification is a good alternative for night-time vehicle detection.

• Journal of electromagnetic engineering and science
• /
• v.12 no.1
• /
• pp.70-76
• /
• 2012

We compare the detection probabilities of the time-reversal(TR) detection system and the conventional radar system. The target is assumed to be hidden inside a random medium such as a forest. We propose a TR detection system based on the SAR(Synthetic Aperture Radar) algorithm. Unlike the conventional SAR images, the proposed TR-SAR system has an interesting property. Specifically, the target-related signal components due to the time-reversal refocusing characteristics, as well as some of clutter-related signal components are concentrated at the time-reversal reference point. The remaining clutter-related signal components are scattered around that reference point. In this paper, we model the random media as a collection of point scatterers to avoid unnecessary complexities. We calculate the detection probability of the TR radar system based on the proposed simple random media model.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.43 no.2
• /
• pp.33-38
• /
• 2020

Anomaly detection of Machine Learning such as PCA anomaly detection and CNN image classification has been focused on cross-sectional data. In this paper, two approaches has been suggested to apply ML techniques for identifying the failure time of big time series data. PCA anomaly detection to identify time rows as normal or abnormal was suggested by converting subjects identification problem to time domain. CNN image classification was suggested to identify the failure time by re-structuring of time series data, which computed the correlation matrix of one minute data and converted to tiff image format. Also, LASSO, one of feature selection methods, was applied to select the most affecting variables which could identify the failure status. For the empirical study, time series data was collected in seconds from a power generator of 214 components for 25 minutes including 20 minutes before the failure time. The failure time was predicted and detected 9 minutes 17 seconds before the failure time by PCA anomaly detection, but was not detected by the combination of LASSO and PCA because the target variable was binary variable which was assigned on the base of the failure time. CNN image classification with the train data of 10 normal status image and 5 failure status images detected just one minute before.

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.14 no.3
• /
• pp.39-44
• /
• 2018

Communication in real-time distributed system should have high reliability. To develop group communication Protocol with high reliability, potential fault should be known and when fault occurs, it should be detected and a necessary action should be taken. Existing detection method by Ack and Time-out is not proper for real time system due to load to Ack which is not received. Therefore, group communication messages from real-time distributed processing systems should be communicated to all receiving processors or ignored by the message itself. This paper can make be sure of transmission of reliable message and deadline by suggesting and experimenting fault detection technique applicable in the real time distributed system based on ring, and analyzing its results. The experiment showed that the shorter the cycle of the heartbeat signal, the shorter the time to propagate the fault detection, which is the time for other nodes to detect the failure of the node.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.8 no.3
• /
• pp.137-143
• /
• 2013

This paper describes an implementation of face detection technology for a real-time application on the Android platform. Java class of Face-Detection for detection of human face is provided by the Android API. However, this function is not suitable to apply for the real-time applications due to inadequate detection speed and accuracy. In this paper, the AdaBoost based classification method which utilizes Local Binary Pattern (LBP) histogram is employed for face detection. The face detection module has been developed by C/C++ language for high-speed image processing, and this module is included to the Android platform using the Java Native Interface (JNI). The experiments were carried out in the Java-based environment and JNI-based environment. The experimental results have shown that the performance of JNI-based is faster than Java-based method and our system is well enough to apply for real-time applications.

• Journal of the Semiconductor & Display Technology
• /
• v.21 no.3
• /
• pp.63-68
• /
• 2022

In this study, we produced defect detection models using parallel convolution neural networks. If convolution neural networks are constructed parallel type, the model's detection accuracy will increase and detection time will decrease. We produced parallel-type defect detection models using 4 types of convolutional algorithms. The performance of models was evaluated using evaluation indicators. The model's performance is detection accuracy and detection time. We compared the performance of each parallel model. The detection accuracy of the model using AlexNet is 97 % and the detection time is 0.3 seconds. We confirmed that when AlexNet algorithm is constructed parallel type, the model has the highest performance.

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.11
• /
• pp.3001-3010
• /
• 1999

The most of intrusion detection methods do not detect intrusion on real-time because it takes a long time to analyze an auditing data for intrusions. To solve the problem, we are studying a real-time intrusion detection. Therefore, this paper proposes an agent model using multi warning level for real-time intrusion detection. It applies to distributed environment using an extensibility and communication mechanism among agents, supports a portability, an extensibility and a confidentiality of IDS.

• Journal of Information Processing Systems
• /
• v.12 no.4
• /
• pp.644-660
• /
• 2016

The real-time detection of malware remains an open issue, since most of the existing approaches for malware categorization focus on improving the accuracy rather than the detection time. Therefore, finding a proper balance between these two characteristics is very important, especially for such sensitive systems. In this paper, we present a fast portable executable (PE) malware detection system, which is based on the analysis of the set of Application Programming Interfaces (APIs) called by a program and some technical PE features (TPFs). We used an efficient feature selection method, which first selects the most relevant APIs and TPFs using the chi-square ($KHI^2$) measure, and then the Phi (${\varphi}$) coefficient was used to classify the features in different subsets, based on their relevance. We evaluated our method using different classifiers trained on different combinations of feature subsets. We obtained very satisfying results with more than 98% accuracy. Our system is adequate for real-time detection since it is able to categorize a file (Malware or Benign) in 0.09 seconds.

• Journal of IKEEE
• /
• v.7 no.2 s.13
• /
• pp.253-259
• /
• 2003

In a good 3D engine, objects interactions are similar to those of real-world. Collision is one of the interactions. It includes whether collision took place or not, where collision took placed, and reaction after collision took place. More precise collision detection needs more time. If there exist required precision, detection time can be controlled by choosing appropriate detection model. Therefore, we need a selection mechanism for the collision detection with respect to required precision and detection time. In this paper, a collision detection model with seven different precision levels is examined. And relationship between detection time and precision is analyzed. Consequently, we propose a selection mechanism for collision detection model.

• International Journal of Internet, Broadcasting and Communication
• /
• v.11 no.4
• /
• pp.76-85
• /
• 2019

We introduce a vision-based object detection method for real-time video surveillance system in low-end edge computing environments. Recently, the accuracy of object detection has been improved due to the performance of approaches based on deep learning algorithm such as Region Convolutional Neural Network(R-CNN) which has two stage for inferencing. On the other hand, one stage detection algorithms such as single-shot detection (SSD) and you only look once (YOLO) have been developed at the expense of some accuracy and can be used for real-time systems. However, high-performance hardware such as General-Purpose computing on Graphics Processing Unit(GPGPU) is required to still achieve excellent object detection performance and speed. To address hardware requirement that is burdensome to low-end edge computing environments, We propose sub-frame analysis method for the object detection. In specific, We divide a whole image frame into smaller ones then inference them on Convolutional Neural Network (CNN) based image detection network, which is much faster than conventional network designed forfull frame image. We reduced its computationalrequirementsignificantly without losing throughput and object detection accuracy with the proposed method.