• Proceedings of the IEEK Conference
• /
• 2003.07e
• /
• pp.1743-1746
• /
• 2003

C. elegans often used to study of function of gene, but it is difficult for human observation to distinguish the mutants of C. elegans. To solve this problem, the system, which can be classified automatically using the computer vision, is studying now. In the previous works , they described the auto-tracking system and the egg-laying timing modeling, which are used to automated-classily system. In this paper, we use three kinds of features, which are related to movement , size and posture of the worm, and each feature is described mathematically and normalized. In experimental result, we validated the features for the hierarchical clustering, And we used the Calinski and Harabasz's method to find the appropriate cluster number.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.8
• /
• pp.3790-3803
• /
• 2018

Content-based image retrieval is an approach used to query images based on their semantics. Semantic based retrieval has its application in all fields including medicine, space, computing etc. Semantically generated binary hash codes can improve content-based image retrieval. These semantic labels / binary hash codes can be generated from unlabeled data using convolutional autoencoders. Proposed approach uses semi-supervised deep hashing with semantic learning and binary code generation by minimizing the objective function. Convolutional autoencoders are basis to extract semantic features due to its property of image generation from low level semantic representations. These representations of images are more effective than simple feature extraction and can preserve better semantic information. Proposed activation and loss functions helped to minimize classification error and produce better hash codes. Most widely used datasets have been used for verification of this approach that outperforms the existing methods.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1997.11a
• /
• pp.228-233
• /
• 1997

An efficient corner matching algorithm is developed to minimize the amount of calculation. To reduce the amount of calculation, all available information from a corner detector is used to make model. This information has uncertainties due to discretization noise and geometric distortion, and this is represented by fuzzy rule base which can represent and handle the uncertainties. Form fuzzy inference procedure, a matched segment list is extracted, and resulted segment list is used to calculate the transformation between object of model and scene. To reduce the false hypotheses, a vote and re-vote method is developed. Also an auto tuning scheme of the fuzzy rule base is developed to find out the uncertainties of features from recognized results automatically. To show the effectiveness of the developed algorithm, experiments are conducted for images of real electronic components.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2003.06a
• /
• pp.18-21
• /
• 2003

In general it has been considered to be the difficult problem that we divide continuous speech into short interval with having identical phoneme quality. In this paper we used Gaussian Mixture Model (GMM) related to probability density to divide speech into phonemes, an initial, medial, and final sound. From them we peformed continuous speech recognition. Decision boundary of phonemes is determined by algorithm with maximum frequency in a short interval. Recognition process is performed by Continuous Hidden Markov Model(CHMM), and we compared it with another phoneme divided by eye-measurement. For the experiments result we confirmed that the method we presented is relatively superior in auto-segmentation in korean speech.

• Proceedings of the KIEE Conference
• /
• 2005.07c
• /
• pp.2208-2210
• /
• 2005

전문적인 지식이 없는 UHF 부분방전 측정장치 사용자를 위해 자동으로 측정된 신호로부터 GIS 내부의 결함을 추론할 수 있는 신경망회로 엔진을 연구하였다. 측정된 방전신호로부터 적절한 변수들을 계산하고 이를 신경망회로를 이용하여 미리 분류한 GIS 결함들 중 가장 유사한 결함을 자동으로 표현하는 기능을 엔진이 가지도록 하였다. 특히 본 엔진은 3상 일괄형 GIS나 GIS의 전압 위상에 동기되지 않은 부분방전 측정시스템에도 방전 원인을 잘 추론함을 실험을 통하여 확인하였다.

• Smart Structures and Systems
• /
• v.30 no.3
• /
• pp.263-271
• /
• 2022

In this paper we present an overview of damage diagnosis algorithms that have been developed over the past two decades using vibration signals obtained from structures. Then, the paper focuses primarily on algorithms that can be used following an extreme event such as a large earthquake to identify structural damage for responding in a timely manner. The algorithms presented in the paper use measurements obtained from accelerometers and gyroscope to identify the occurrence of damage and classify the damage. Example algorithms are presented include those based on autoregressive moving average (ARMA), wavelet energies from wavelet transform and rotation models. The algorithms are illustrated through application of data from test structures such as the ASCE Benchmark structure and laboratory tests of scaled bridge columns and steel frames. The paper concludes by identifying needs for research and development in order for such algorithms to become viable in practice.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.3
• /
• pp.101-108
• /
• 2023

Optimizing the performance of deep neural networks on embedded systems is a challenging task that requires efficient compilers and runtime systems. We propose a TVM-based approach that consists of three steps: quantization, auto-scheduling, and ahead-of-time compilation. Our approach reduces the computational complexity of models without significant loss of accuracy, and generates optimized code for various hardware platforms. We evaluate our approach on three representative CNNs using ImageNet Dataset on the NVIDIA Jetson AGX Xavier board and show that it outperforms baseline methods in terms of processing speed.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.39C no.10
• /
• pp.887-895
• /
• 2014

Drowsy driving is a large proportion of the total car accidents. For this reason, drowsiness detection and warning system for drivers has recently become a very important issue. Monitoring physiological signals provides the possibility of detecting features of drowsiness and fatigue of drivers. Many researches have been published that to measure electroencephalogram(EEG) signals is the effective way in order to be aware of fatigue and drowsiness of drivers. The aim of this study is to extract drowsiness-related features from a set of EEG signals and to classify the features into three states: alertness, transition, and drowsiness. This paper proposes a drowsiness detection system using errors-in-variables(EIV) for extraction of feature vectors and multilayer perceptron (MLP) for classification. The proposed method evaluates robustness for noise and compares to the previous one using linear predictive coding (LPC) combined with MLP. From evaluation results, we conclude that the proposed scheme outperforms the previous one in the low signal-to-noise ratio regime.

• Journal of Auto-vehicle Safety Association
• /
• v.10 no.1
• /
• pp.20-26
• /
• 2018

The fatality of rollover accidents in motor vehicle crashes is high despite their low incidence. Through the investigation of a 12-passenger van rollover accident in which 10 passengers were involved, we intend to analyze the correlation between the severity of the injury and the position of the occupants. We collected accident information from medical records, interviews, photo-images of the damaged van, field surveys, and the results of the Korean New Car Assessment Program (KNCAP). Based on the occupants' position, we classified injury sites and estimated injury severity. Passenger injury severity was evaluated by trauma score calculation. The initiation type of the rollover accident was passenger side 'fall-over' and the Collision Deformation Classification (CDC) code for the damaged van was 00TDZO3. The crash of the van involved 10 passengers, with an average age of $16.3{\pm}4.2years$. Few of the occupants had fastened seat belts at the time of the incident, and there was no airbag installed. One patient sustained severe liver injury and another was diagnosed with a fracture of the right humerus. The most common injuries were at the upper extremities and the neck. The average of Injury Severity Score (ISS) was $4.8{\pm}5.9$, and the average ISS of right-seated, mid-seated and left-seated occupants was $7.5{\pm}9.3$, $1.5{\pm}0.7$, and $3.3{\pm}2.1$ respectively (p>0.05). In the rollover (to-passenger side) accident of occupant unfastened, the average ISS of right-seated occupants (near side) was higher, but there was no statistically significant difference.

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.10
• /
• pp.1010-1014
• /
• 2010

Article-related product recommender system is an emerging e-commerce service which recommends items based on association in contexts between items and articles. Current services recommend based on the similarity between tags of articles and items, which is deficient not only due to the high cost in manual tagging but also low accuracies in recommendation. As a component of novel article-related item recommender system, we propose a new method for tagging item images based on pre-defined categories. We suggest a hypernetwork-based algorithm for learning association between images, which is represented by visual words, and categories of products. Learned hypernetwork are used to assign multiple tags to unlabeled item images. We show the ability of our method with a product set of real-world online shopping-mall including 1,251 product images with 10 categories. Experimental results not only show that the proposed method has competitive tagging performance compared with other classifiers but also present that the proposed multi-tagging method based on hypernetworks improves the accuracy of tagging.