• Journal of Satellite, Information and Communications
• /
• v.9 no.4
• /
• pp.77-81
• /
• 2014

This paper propose a method to detect moving object and locating in real time from video sequence. first the proposed method extract moving object by differencing two consecutive frames from the video sequence. If the interval between captured two frames is long, it cause to generate fake moving object as tail of the real moving object. secondly this paper proposed method to overcome this problem by using doppler effects and HSV color model. finally the object segmentation and locating is done by combining the result that obtained from steps above. The proposed method has 99.2% of detection rate in practical and also this method is comparatively speed than other similar methods those proposed in past. Since the complexity of the algorithm is directly affects to the speed of the system, the proposed method can be used as low complexity algorithm for real time moving object detection.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.20 no.4
• /
• pp.516-521
• /
• 2010

In this paper, we propose a video-based intelligent unmanned fire surveillance system using fuzzy color models. In general, to detect heat or smoke, a separate device is required for a fire surveillance system, this system, however, can be implemented by using widely used CCTV, which does not need separate devices and extra cost. The systems called video-based fire surveillance systems use mainly a method extracting smoke or flame from an input image only. The smoke is difficult to extract at night because of its gray-scale color, and the flame color depends on the temperature, the inflammable, the size of flame, etc, which makes it hard to extract the flame region from the input image. This paper deals with a intelligent fire surveillance system which is robust against the variation of the flame color, especially at night. The proposed system extracts the moving object from the input image, makes a decision whether the object is the flame or not by means of the color obtained by fuzzy color model and the shape obtained by histogram, and issues a fire alarm when the flame is spread. Finally, we verify the efficiency of the proposed system through the experiment of the controlled real fire.

• Journal of Broadcast Engineering
• /
• v.26 no.6
• /
• pp.809-812
• /
• 2021

Air bubbles which may be generated during the PCB coating process can be a major cause of malfunction. so it is necessary to detect the bubbles in advance. In previous studies, candidates for bubbles were extracted using the brightness characteristics of bubbles, and the candidates were verified using CNN(Convolutional Neural Networks). In this paper, we propose a bubble detection method using a transfer learning-based CNN model. The VGGNet is adopted and sigmoid is used as a classification layer, and the last convolutional layer and classification layer are trained together when transfer learning is applied. The performance of the proposed method is F1-score 0.9044, which shows an improvement of about 0.17 compared to the previous study.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.5
• /
• pp.414-423
• /
• 2020

In this paper, we propose a Dilated Convolution Gate Linear Unit (DCGLU) to mitigate the lack of sparsity and small receptive field problems caused by the segmentation map extraction process in sound event detection with weak labels. In the advent of deep learning framework, segmentation map extraction approaches have shown improved performance in noisy environments. However, these methods are forced to maintain the size of the feature map to extract the segmentation map as the model would be constructed without a pooling operation. As a result, the performance of these methods is deteriorated with a lack of sparsity and a small receptive field. To mitigate these problems, we utilize GLU to control the flow of information and Dilated Convolutional Neural Networks (DCNNs) to increase the receptive field without additional learning parameters. For the performance evaluation, we employ a URBAN-SED and self-organized bird sound dataset. The relevant experiments show that our proposed DCGLU model outperforms over other baselines. In particular, our method is shown to exhibit robustness against nature sound noises with three Signal to Noise Ratio (SNR) levels (20 dB, 10 dB and 0 dB).

• KIPS Transactions on Computer and Communication Systems
• /
• v.11 no.10
• /
• pp.323-332
• /
• 2022

As various services using AI technology are being developed, much attention is being paid to AI service production. Recently, AI technology is acknowledged as one of ICT services, a lot of research is being conducted for general-purpose AI service production. In this paper, I describe the research results in terms of systems for AI service production, focusing on the distribution and production of machine learning models, which are the final steps of general machine learning development procedures. Three different Ubuntu systems were built, and experiments were conducted on the system, using data from 2017 validation COCO dataset in combination of different AI models (RFCN, SSD-Mobilenet) and different communication methods (gRPC, REST) to request and perform AI services through Tensorflow serving. Through various experiments, it was found that the type of AI model has a greater influence on AI service inference time than AI machine communication method, and in the case of object detection AI service, the number and complexity of objects in the image are more affected than the file size of the image to be detected. In addition, it was confirmed that if the AI service is performed remotely rather than locally, even if it is a machine with good performance, it takes more time to infer the AI service than if it is performed locally. Through the results of this study, it is expected that system design suitable for service goals, AI model development, and efficient AI service production will be possible.

• The Journal of Bigdata
• /
• v.7 no.2
• /
• pp.31-41
• /
• 2022

Steel is one of the most fundamental components to mechanical industry. However, the quality of products are greatly impacted by the surface defects in the steel. Thus, researchers pay attention to the need for surface defects detector and the deep learning methods are the current trend of object detector. There are still limitations and rooms for improvements, for example, related works focus on developing the models but don't take into account real-time application with practical implication on industrial settings. In this paper, a real-time application of steel surface defects detection based on YOLOv4 is proposed. Firstly, as the aim of this work to deploying model on real-time application, we studied related works on this field, particularly focusing on one-stage detector and YOLO algorithm, which is one of the most famous algorithm for real-time object detectors. Secondly, using pre-trained Yolov4-Darknet platform models and transfer learning, we trained and test on the hot rolled steel defects open-source dataset NEU-DET. In our study, we applied our application with 4 types of typical defects of a steel surface, namely patches, pitted surface, inclusion and scratches. Thirdly, we evaluated YOLOv4 trained model real-time performance to deploying our system with accuracy of 87.1 % mAP@0.5 and over 60 fps with GPU processing.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.8
• /
• pp.1-8
• /
• 2020

Air ducts installed for ventilation inside buildings accumulate contaminants during their service life. Robots are installed to clean the air duct at low cost, but they are still not fully automated and depend on manpower. In this study, an intersection detection algorithm for autonomous driving was applied to an air duct cleaning robot. Autonomous driving of the robot was achieved by calculating the distance and angle between the extracted point and the center point through the intersection detection algorithm from the camera image mounted on the robot. The training data consisted of CAD images of the duct interior as well as the cross-point coordinates and angles between the two boundary lines. The deep learning-based CNN model was applied as a detection algorithm. For training, the cross-point coordinates were obtained from CAD images. The accuracy was determined based on the differences in the actual and predicted areas and distances. A cleaning robot prototype was designed, consisting of a frame, a Raspberry Pi computer, a control unit and a drive unit. The algorithm was validated by video imagery of the robot in operation. The algorithm can be applied to vehicles operating in similar environments.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.44 no.2
• /
• pp.46-54
• /
• 2007

In this paper, we introduce the performance of improved symbol timing detection by the superposed difference type symbol timing detection method in the OFDM system. Also, we represent the maximum detection probability of symbol synchronization timing at each received delay signal in multipath channel delay profile in the correlation and difference type symbol timing detection methods. The computer simulation results show that the correlation symbol timing detection method have maximum detection probability at the lead of the nth received delay signal of highest amplitude, but the difference type symbol timing detection method always have maximum detection probability at the lead of the first received delay signal in the channel delay spread of $70nsec{\sim}217nsec$. The simulation results indicate the possibility of the perfect detection of OFDM symbol synchronization timing and it fit well with the results of improved S/N to the eb/n0 and the performance of symbol timing detection of the proposed method.

• Journal of Internet of Things and Convergence
• /
• v.6 no.1
• /
• pp.83-95
• /
• 2020

Being a dense and complex task of precisely detecting the weeds in practical crop field environment, previous approaches lack in terms of speed of processing image frames with accuracy. Although much of the attention has been given to classify the plants diseases but detecting crop weed issue remained in limelight. Previous approaches report to use fast algorithms but inference time is not even closer to real time, making them impractical solutions to be used in uncontrolled conditions. Therefore, we propose a detection model for the complex rice weed detection task. Experimental results show that inference time in our approach is reduced with a significant margin in weed detection task, making it practically deployable application in real conditions. The samples are collected at two different growth stages of rice and annotated manually

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.10 no.5
• /
• pp.435-441
• /
• 2017

Detecting an inter-harmonic accurately is not easy work, because it has small magnitude, and its frequency which can be observed is not an integer multiple of fundamental frequency. In this paper, a new method using filter bank system and adaptive predictor is proposed. Filter bank system decomposes input signal to sub bands. In adaptive predictor, inter-harmonic is detected with decomposed sub band signal as input, and error signal as output. In this scheme, input-output characteristic of adaptive predictor is notch filter, as predicted harmonic is canceled in error signal, so detecting an inter-harmonic can be possible. Magnitude and frequency of detected inter-harmonic is estimated by recursive algorithm. The performances of proposed method are evaluated to sinusoidal signal model synthesized with harmonics and inter-harmonics. And validity of the method is proved as comparing the inter-harmonic detection results to MUSIC and ESPRIT.