• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.01a
• /
• pp.592-595
• /
• 2009

In this paper, we propose a face detection technique for still pictures which sequentially uses a skin-color model and a support vector machine (SVM). SVM is a learning algorithm for solving the classification problem. Some studies on face detection have reported superior results of SVM over neural networks. The SVM method searches for a face in a picture while changing the size of the window. The detection accuracy and the processing time of SVM vary largely depending on the complexity of the background of the picture or the size of the face. Therefore, we apply a face candidate area detection method using a skin-color model as a preprocessing technique. We compared the method using SVM alone with that of the proposed method in respect to face detection accuracy and processing time. As a result, the proposed method showed improved processing time while maintaining a high recognition rate.

• Journal of Communications and Networks
• /
• v.15 no.3
• /
• pp.276-282
• /
• 2013

Self-encoded spread spectrum (SESS) is a novel modulation technique that acquires its spreading code from a random information source, rather than using the traditional pseudo-random noise (PN) codes. In this paper, we present our study of the SESS system performance under pulsed-noise jamming and show that iterative detection can significantly improve the bit error rate (BER) performance. The jamming performance of the SESS with correlation detection is verified to be similar to that of the conventional direct sequence spread spectrum (DSSS) system. On the other hand, the time diversity detection of the SESS can completely mitigate the effect of jamming by exploiting the inherent temporal diversity of the SESS system. Furthermore, iterative detection with multiple iterations can not only eliminate the jamming completely but also achieve a gain of approximately 1 dB at $10^{-3}$ BER as compared with the binary phase shift keying (BPSK) system under additive white gaussian noise (AWGN) by effectively combining the correlation and time diversity detections.

• Korean Journal of Remote Sensing
• /
• v.37 no.4
• /
• pp.747-761
• /
• 2021

With the increasing application of UAVs in intelligent transportation systems, vehicle detection for aerial images has become an essential engineering technology and has academic research significance. In this paper, a vehicle detection method for aerial images based on the YOLOv4 deep learning algorithm is presented. At present, the most known datasets are VOC (The PASCAL Visual Object Classes Challenge), ImageNet, and COCO (Microsoft Common Objects in Context), which comply with the vehicle detection from UAV. An integrated dataset not only reflects its quantity and photo quality but also its diversity which affects the detection accuracy. The method integrates three public aerial image datasets VAID, UAVD, DOTA suitable for YOLOv4. The training model presents good test results especially for small objects, rotating objects, as well as compact and dense objects, and meets the real-time detection requirements. For future work, we will integrate one more aerial image dataset acquired by our lab to increase the number and diversity of training samples, at the same time, while meeting the real-time requirements.

• Journal of Information Processing Systems
• /
• v.18 no.3
• /
• pp.428-442
• /
• 2022

A reliable automatic passenger counting (APC) system is a key point in transportation related to the efficient scheduling and management of transport routes. In this study, we introduce a lightweight head detection network using deep learning applicable to an embedded system. Currently, object detection algorithms using deep learning have been found to be successful. However, these algorithms essentially need a graphics processing unit (GPU) to make them performable in real-time. So, we modify a Tiny-YOLOv3 network using certain techniques to speed up the proposed network and to make it more accurate in a non-GPU environment. Finally, we introduce an APC system, which is performable in real-time on embedded systems, using the proposed head detection algorithm. We implement and test the proposed APC system on a Samsung ARTIK 710 board. The experimental results on three public head datasets reflect the detection accuracy and efficiency of the proposed head detection network against Tiny-YOLOv3. Moreover, to test the proposed APC system, we measured the accuracy and recognition speed by repeating 50 instances of entering and 50 instances of exiting. These experimental results showed 99% accuracy and a 0.041-second recognition speed despite the fact that only the CPU was used.

• Korean Journal of Artificial Intelligence
• /
• v.11 no.3
• /
• pp.17-22
• /
• 2023

In this paper, we propose a black ice detection system using Convolutional Neural Networks (CNNs). Black ice poses a serious threat to road safety, particularly during winter conditions. To overcome this problem, we introduce a CNN-based architecture for real-time black ice detection with an encoder-decoder network, specifically designed for real-time black ice detection using thermal images. To train the network, we establish a specialized experimental platform to capture thermal images of various black ice formations on diverse road surfaces, including cement and asphalt. This enables us to curate a comprehensive dataset of thermal road black ice images for a training and evaluation purpose. Additionally, in order to enhance the accuracy of black ice detection, we propose a multi-scale dilation convolution feature fusion (MsDC-FF) technique. This proposed technique dynamically adjusts the dilation ratios based on the input image's resolution, improving the network's ability to capture fine-grained details. Experimental results demonstrate the superior performance of our proposed network model compared to conventional image segmentation models. Our model achieved an mIoU of 95.93%, while LinkNet achieved an mIoU of 95.39%. Therefore, it is concluded that the proposed model in this paper could offer a promising solution for real-time black ice detection, thereby enhancing road safety during winter conditions.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.27 no.4
• /
• pp.311-323
• /
• 2023

This paper introduces the industrial problem, the solution, and the results of the research conducted with Define Inc. The client company wanted to improve the performance of an object detection model on the floor plan dataset. To solve the problem, we analyzed the operational principles, advantages, and disadvantages of the existing object detection model, identified the characteristics of the floor plan dataset, and proposed to use of YOLO v5 as an appropriate object detection model for training the dataset. We compared the performance of the existing model and the proposed model using mAP@60, and verified the object detection results with real test data, and found that the performance increase of mAP@60 was 0.08 higher with a 25% shorter inference time. We also found that the training time of the proposed YOLO v5 was 71% shorter than the existing model because it has a simpler structure. In this paper, we have shown that the object detection model for the floor plan dataset can achieve better performance while reducing the training time. We expect that it will be useful for solving other industrial problems related to object detection in the future. We also believe that this result can be extended to study object recognition in 3D floor plan dataset.

• Proceedings of the KIEE Conference
• /
• 2008.04a
• /
• pp.159-160
• /
• 2008

Initial Ranging Detection and Synchronization is suggested for IEEE 802.16e OFDMA Systems. However ranging is vulnerable to the channel selectivity and other user's interference at low SNR. This paper presents enhanced ranging scheme that improves ranging detection process using the combine multiple FFT blocks and cope with channel selectivity and other user's interference at low SNR. Based on the ranging detection timing offset is calculated for synchronization.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.35 no.2
• /
• pp.71-88
• /
• 2010

A density-based outlier detection such as an LOF (Local Outlier Factor) tries to find an outlying observation by using density of its surrounding space. In spite of several advantages of a density-based outlier detection method, the computational complexity of outlier detection has been one of major barriers in its application. In this paper, we present an LOF algorithm that can reduce computation time of a density based outlier detection algorithm. A kd-tree indexing and approximated k-nearest neighbor search algorithm (ANN) are adopted in the proposed method. A set of experiments was conducted to examine performance of the proposed algorithm. The results show that the proposed method can effectively detect local outliers in reduced computation time.

• Journal of Korea Multimedia Society
• /
• v.7 no.12
• /
• pp.1692-1699
• /
• 2004

Many researchers have developed various techniques for detection of human faces in ordinary still images. Face detection is the first imperative step of human face recognition systems. The two main problems of human face detection are how to cutoff the running time and how to reduce the number of false positives. In this paper, we present frontal and near-frontal face detection algorithm in still gray images using a thick edge image and neural network. We have devised a new filter that gets the thick edge image. Our overall scheme for face detection consists of two main phases. In the first phase we describe how to create the thick edge image using the filter and search for face candidates using a whole face detector. It is very helpful in removing plenty of windows with non-faces. The second phase verifies for detecting human faces using component-based eye detectors and the whole face detector. The experimental results show that our algorithm can reduce the running time and the number of false positives.

• The Transactions of the Korean Institute of Electrical Engineers P
• /
• v.65 no.1
• /
• pp.66-71
• /
• 2016

According to the development of depth sensing devices and depth estimation technology, depth information becomes more important for object detection in computer vision. In terms of recognition rate, pedestrian detection methods have been improved more accurately. However, the methods makes slower detection time. So, many researches have overcome this problem by using GPU. Here, we propose a real-time pedestrian detection algorithm that does not rely on GPU. First, the depth-weighted distance map is used for detecting expected human regions. Next, human detection is performed on the regions. The performance for the proposed approach is evaluated and compared with the previous methods. We show that proposed method can detect human about 7 times faster than conventional ones.