• Journal of Korea Multimedia Society
• /
• v.23 no.8
• /
• pp.915-926
• /
• 2020

Outlier detection methods help one to identify unusual instances in data that may correspond to erroneous, exceptional, or surprising events or behaviors. This work studies conditional outlier detection, a special instance of the outlier detection problem, in the context of incorrect data label identification. Unlike conventional (unconditional) outlier detection methods that seek abnormalities across all data attributes, conditional outlier detection assumes data are given in pairs of input (condition) and output (response or label). Accordingly, the goal of conditional outlier detection is to identify incorrect or unusual output assignments considering their input as condition. As a solution to conditional outlier detection, this paper proposes the ratio-based outlier scoring (ROS) approach and its variant. The propose solutions work by adopting conventional outlier scores and are able to apply them to identify conditional outliers in data. Experiments on synthetic and real-world image datasets are conducted to demonstrate the benefits and advantages of the proposed approaches.

• International journal of advanced smart convergence
• /
• v.10 no.1
• /
• pp.82-88
• /
• 2021

In this paper, we propose an automatic detection scheme of work distraction for remote management of telecommuting. The proposed scheme periodically captures two consequent computer screens and generates the difference image of these two captured images. The scheme applies the difference image to our deep learning model and makes a decision of abnormal patterns in the difference image. Our deep learning model is designed with the transfer learning technique of VGG16 deep learning. When the scheme detects an abnormal pattern in the difference image, it hides all texts in the difference images to protect disclosure of privacy-related information. Evaluation shows that the proposed scheme provides about 96% detection accuracy.

• Proceedings of the KIEE Conference
• /
• 2007.04a
• /
• pp.426-429
• /
• 2007

When a fire happens in builds or apartments, peoples are tend to be caught in dangerous situations owing to the failure of searching escape route to the outside. In this work, an efficient fire detection and alarm system which makes it possible for the escapers to take adequate actions is proposed. The proposed system consists of two parts. One is fire detection modules which are located at each compartments in a building. The other is fire warning modules equipped with portable flashes having ability of visual/voice warning. Fire detection information is transmitted between each modules wirelessly. In this work, an efficient communication protocol for sensor network-based fire detection system is proposed and its feasibility is verified by practical experiments using MANGO ZDK Development tools

• Journal of Korean Society of Occupational and Environmental Hygiene
• /
• v.28 no.1
• /
• pp.100-107
• /
• 2018

Objectives: This study examines exposure to hazardous substances in the working environment caused by exposure to toxic substances produced in the aluminum die casting process in the automobile manufacturing industry. Materials and Methods: The exposure concentration levels, detection rates and time-trend of 15 hazardous factors in the aluminum die casting process over 10 years(from 2006 to 2016) were used as a database. Results: The study found that hazardous factors in the aluminum die casting process were mostly metals. The rate for detected samples was 70.6%(405 samples), and that for not detected samples was 29.4%. The noise for an eight-hour work shift showed a 49.7% exceedance rate for TLV-TWA. Average noise exposure was 89.0 dB. The maximum exposure level was 105.1 dB. Conclusion: The high numbers of no-detection rates for hazardous substance exposure shows that there is no need to do a work environment measurement. Therefore, alternatives are necessary for improving the efficiency and reliability of the work environment measurement. Moreover, to prevent noise damage, reducing noise sources from automation, shielding, or sound absorbents are necessary.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.5
• /
• pp.1597-1610
• /
• 2022

At present, the main method of high-speed train chassis detection is using computer vision technology to extract keypoints from two related chassis images firstly, then matching these keypoints to find the pixel-level correspondence between these two images, finally, detection and other steps are performed. The quality and accuracy of image matching are very important for subsequent defect detection. Current traditional matching methods are difficult to meet the actual requirements for the generalization of complex scenes such as weather, illumination, and seasonal changes. Therefore, it is of great significance to study the high-speed train image matching method based on deep learning. This paper establishes a high-speed train chassis image matching dataset, including random perspective changes and optical distortion, to simulate the changes in the actual working environment of the high-speed rail system as much as possible. This work designs a convolutional neural network to intensively extract keypoints, so as to alleviate the problems of current methods. With multi-level features, on the one hand, the network restores low-level details, thereby improving the localization accuracy of keypoints, on the other hand, the network can generate robust keypoint descriptors. Detailed experiments show the huge improvement of the proposed network over traditional methods.

• Journal of Multimedia Information System
• /
• v.1 no.1
• /
• pp.55-67
• /
• 2014

Image topic and emotion analysis is an important component of online image retrieval, which nowadays has become very popular in the widely growing social media community. However, due to the gaps between images and texts, there is very limited work in literature to detect one image's Topics and Emotions in a unified framework, although topics and emotions are two levels of semantics that often work together to comprehensively describe one image. In this work, a unified model, Joint Topic/Emotion Multi-Modal Hierarchical Latent Dirichlet Allocation (JTE-MMHLDA) model, which extends previous LDA, mmLDA, and JST model to capture topic and emotion information at the same time from heterogeneous data, is proposed. Specifically, a two level graphical structured model is built to realize sharing topics and emotions among the whole document collection. The experimental results on a Flickr dataset indicate that the proposed model efficiently discovers images' topics and emotions, and significantly outperform the text-only system by 4.4%, vision-only system by 18.1% in topic detection, and outperforms the text-only system by 7.1%, vision-only system by 39.7% in emotion detection.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.12
• /
• pp.4816-4834
• /
• 2020

This paper proposes transfer learning and fine-tuning techniques for a deep learning model to detect three distinct brain tumors from Magnetic Resonance Imaging (MRI) scans. In this work, the recent YOLOv4 model trained using a collection of 3064 T1-weighted Contrast-Enhanced (CE)-MRI scans that were pre-processed and labeled for the task. This work trained with the partial 29-layer YOLOv4-Tiny and fine-tuned to work optimally and run efficiently in most platforms with reliable performance. With the help of transfer learning, the model had initial leverage to train faster with pre-trained weights from the COCO dataset, generating a robust set of features required for brain tumor detection. The results yielded the highest mean average precision of 93.14%, a 90.34% precision, 88.58% recall, and 89.45% F1-Score outperforming other previous versions of the YOLO detection models and other studies that used bounding box detections for the same task like Faster R-CNN. As concluded, the YOLOv4-Tiny can work efficiently to detect brain tumors automatically at a rapid phase with the help of proper fine-tuning and transfer learning. This work contributes mainly to assist medical experts in the diagnostic process of brain tumors.

• Journal of Navigation and Port Research
• /
• v.32 no.2
• /
• pp.133-140
• /
• 2008

In this paper, a fault detection method for operator failures using the observation technique is proposed. The suggested algorithm is extended using the conventional sensor/actuator fault detection method. First, it is assumed that operator failure affects human work operations, as it is an external input signal. With this assumption, a human work model with operator failure is suggested. Second, an unknown input observer with proportional and integral gains is introduced. The characteristic of this observer of estimating an external signal without an exact input is shown, and the conditions for the detection of an operator failure are proposed. Finally, by simulating the container crane operations, it is verified that the observer can accurately detect an operator failure and estimate its magnitude from the given internal signal.

• International Journal of Internet, Broadcasting and Communication
• /
• v.13 no.1
• /
• pp.238-242
• /
• 2021

Game theory has been regarded as a useful theoretical tool for modeling the interactions between distinct entities and thus it has been harnessed in various research field. In particular, research attention has been shown to how to apply game theory to modeling the interactions between malign and benign entities in the field of wireless networks. Although various game theoretic modeling work have been proposed in the field of wireless networks, our proposed work is disparate to the existing work in the sense that we focus on mobile malign node detection problem in static wireless sensor networks. More specifically, we propose a Bayesian game theoretic modeling for mobile malign node detection problem in static wireless sensor networks. In our modeling, we formulate a two-player static Bayesian game with imperfect information such that player 1 is aware of the type of player 2, but player 2 is not aware of the type of player 1. We use four strategies in our static Bayesian game. We obtain Bayesian Nash Equilibria with pure strategies under certain conditions.

• International Journal of Contents
• /
• v.9 no.3
• /
• pp.9-14
• /
• 2013

In this paper, we will examine single face tracking algorithms with scaling function in a mobile device. Face detection and tracking either in PC or mobile device with scaling function is an unsolved problem. Standard single face tracking method with particle filter has a problem in tracking the objects where the object can move closer or farther from the camera. Therefore, we create an algorithm which can work in a mobile device and perform a scaling function. The key idea of our proposed method is to extract the average of skin color in face detection, then we compare the skin color distribution between the detected face and the tracking face. This method works well if the face position is located in front of the camera. However, this method will not work if the camera moves closer from the initial point of detection. Apart from our weakness of algorithm, we can improve the accuracy of tracking.