• 반도체디스플레이기술학회지
• /
• 제22권2호
• /
• pp.40-44
• /
• 2023

The defect detection based on deep learning requires minimal loss and high accuracy to pinpoint product defects. In this paper, we confirm the loss rate of deep learning training based on disc-shaped artificial scaffold images. It is intended to compare the performance of Cross-Entropy functions used in object detection algorithms. The model was constructed using normal, defective artificial scaffold images and category cross entropy and sparse category cross entropy. The data was repeatedly learned five times using each loss function. The average loss rate, average accuracy, final loss rate, and final accuracy according to the loss function were confirmed.

• 대한전자공학회논문지SP
• /
• 제42권6호
• /
• pp.159-166
• /
• 2005

레이더의 스캔특성에 의해 ES 시스템에 수신된 탐색 레이더 신호의 펄스세기는 펄스마다 일정하지 않다. 이러한 펄스세기의 변화는 ES 탐지손실을 유발하므로 스캔에 따른 탐지손실을 ES 탐지거리 방정식에 고려하여야 한다. 본 논문에서는 원형스캔에 대하여 ES 탐지손실을 이론적으로 분석하고, 정량적으로 예측할 수 있는 모델을 제안하였다. 실제 레이더에 대해 탐지손실을 측정한 결과, 제안된 모델이 원형 스캔에 관계된 ES 탐지손실 모델로 적합함을 알 수 있었다.

• 한국콘텐츠학회논문지
• /
• 제19권9호
• /
• pp.637-645
• /
• 2019

RFID 그룹증명은 다수의 태그가 동시에 스캔 되었음을 증명하는 요킹증명의 확장이다. 기존의 그룹증명 기법들은 태그응답의 손실을 검증단계에서 감지하는 지연된 태그손실 감지를 지원한다. 그러나 지연된 태그손실 감지는 태그의 손실을 즉각적으로 감지해야 하는 실시간 응용에는 적합하지 못하다. 이 연구에서 나는 태그의 손실을 빠르게 감지하는 새로운 태그응답손실 감지기법인 TRLD(Tag Response Loss Detection)를 제안한다. 제안기법에서 태그는 응답과 함께 시퀀스번호를 전송하며, 리더는 시퀀스번호를 통해 태그를 식별하는 과정 없이 태그응답의 손실을 감지한다. 안전성 분석에서는 메시지 비구별성 실험을 통해, 시퀀스번호가 특정태그와 태그그룹을 구분하려고 시도하는 메시지 분석 공격에 대해 안전하다는 것을 보인다. 효율성 측면에서 제안기법은 어떤 태그의 응답이 손실되었는지 확정하기 위해 기존의 기법보다 더 적은 수의 통신과 데이터베이스 연산을 요구한다.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2001년도 봄 학술발표논문집 Vol.28 No.1 (A)
• /
• pp.163-165
• /
• 2001

본 논문에서는 먼저 Forwarding 방식의 라우팅 프로토콜인 MPLS(Multi-Protocol Label Switching)와 네트웍에서의 Traffic Engineering(TE)에 대한 개괄적인 설명과 함께, MPLS 네트웍 내에서의 트래픽 엔지니어링에 대해 기술한다. 그리고 MPLS 도메인 양 끝단에서 단일 경로의 패킷에 대한 MPLS 헤더의 레이블 번호를 이용한 동일한 패킷인지에 대한 확인 방안과 MPLS 도메인 내에서 Loss Detection 메커니즘을 이용한 효율적인 트래픽 엔지니어링방안을 제안한다. 향후 본 연구 방안을 적용하게 되면, 차등 서비스(Differentiated Services, Diffserv)를 제공하는 네트웍 환경의 핵심 망과 Mobile IP 기반의 무선 네트웍 환경에서 유선 네트웍의 Quality of Service(QoS)를 향상시킬 수 있을 것이다.

• 대한간호학회지
• /
• 제41권6호
• /
• pp.758-767
• /
• 2011

Purpose: This study was done to examine the threshold value of estimated height loss at which the risk of osteoporosis increases and to verify its discriminative ability in the detection of osteoporosis. Methods: It was conducted based on epidemiological descriptive methods on 732 Korean women at a public healthcare center in Seoul between July and November 2010. ANOVA, Pearson correlation, logistic regression analysis and receiver operating characteristics (ROC) curve were used for data analysis. Results: There was an age-related correlation between bone mineral density (lumbar spine: F=37.88, p<.001; femur: F=54.27, p<.001) and estimated height loss (F=27.68, p<.001). Estimated height loss increased significantly with decreasing bone mineral density (lumbar spine: r=-.23, p<.001; femur: r=-.34, p<.001). The odds ratio for the point at which the estimated height loss affects the occurrence of osteoporosis was found to increase at a cut-off value of 2 cm and the area under ROC curve was .71 and .82 in lumbar spine and femur, respectively. Conclusion: The optimal cut-off value of the estimated height loss for detection of osteoporosis was 2 cm. Height loss is therefore a useful indicator for the self-assessment and prognosis of osteoporosis.

• Smart Structures and Systems
• /
• 제29권1호
• /
• pp.153-166
• /
• 2022

Identifying fine cracks in steel bridge facilities is a challenging task of structural health monitoring (SHM). This study proposed an end-to-end crack image segmentation framework based on a one-step Convolutional Neural Network (CNN) for pixel-level object recognition with high accuracy. To particularly address the challenges arising from small object detection in complex background, efforts were made in loss function selection aiming at sample imbalance and module modification in order to improve the generalization ability on complicated images. Specifically, loss functions were compared among alternatives including the Binary Cross Entropy (BCE), Focal, Tversky and Dice loss, with the last three specialized for biased sample distribution. Structural modifications with dilated convolution, Spatial Pyramid Pooling (SPP) and Feature Pyramid Network (FPN) were also performed to form a new backbone termed CrackDet. Models of various loss functions and feature extraction modules were trained on crack images and tested on full-scale images collected on steel box girders. The CNN model incorporated the classic U-Net as its backbone, and Dice loss as its loss function achieved the highest mean Intersection-over-Union (mIoU) of 0.7571 on full-scale pictures. In contrast, the best performance on cropped crack images was achieved by integrating CrackDet with Dice loss at a mIoU of 0.7670.

• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제10권3호
• /
• pp.109-114
• /
• 2021

GAN(Generative Adversarial Network, 생성적 적대 신경망)은 이미지 생성모델로서 생성기 네트워크와 판별기 네트워크로 구성되며 실제 같은 이미지를 생성한다. GAN에 의해 생성된 이미지는 실제 이미지와 유사해야 하므로 생성된 이미지와 실제 이미지의 손실 오차를 최소화하는 손실함수(loss function)를 사용한다. 그러나 GAN의 손실함수는 이미지를 생성하는 학습을 불안정하게 만들어 이미지의 품질을 떨어뜨린다는 문제점이 있다. 이러한 문제를 해결하기 위해 본 논문에서는 GAN 관련 연구를 분석하고 에지 검출(edge detection)을 이용한 eGAN(edge GAN)을 제안한다. 실험 결과 eGAN 모델이 기존의 GAN 모델보다 성능이 개선되었다.

• 전기학회논문지
• /
• 제62권12호
• /
• pp.1803-1808
• /
• 2013

To maintain contact between catenary and pantograph copper is important in order to transmit power smoothly on Current collection system. But, Arc discharge with strong light is generated because of contact loss. Therefore, Arc discharge detection is important measurement factor judging performance of current collection system. In this paper, It is described to results of arc discharge applying UV detection technology using arc generator. And Arc discharge was detected using the most commonly used processing catenary and rigid catenary and pantograph copper of electric rolling stock for securing arc detection instrument reliability. Results of contact loss detection instrument in this paper will be used for maintenance of current collection quality and system.

• 한국해양공학회지
• /
• 제8권1호
• /
• pp.62-70
• /
• 1994

In order to estimate the detection range of a active SONAR system, the SONAR equation is commonly used. In this paper, an algorithm to calculate detection range in active SONAR system as function of SONAR depth and target depth is presented. For given SONAR parameters and environment, the transmission loss and background level are found, signal excess is computed. Using log-normal distribution, signal excess is converted to detection probability at each range. Then, the detection range is obtained by integrating the detection probability as function of range for each depth. The proposed algorithm have been applied to the case of omni-directional source with center frequency 30Hz for summer and winter sound profiles. It is found that the optimal search depth is the source depth since the detection range increase at source depth where the signal excess is maximized.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제18권8호
• /
• pp.2199-2213
• /
• 2024

To enhance the quality of defect detection for Printed Circuit Board Assembly (PCBA) during electronic product manufacturing, this study primarily focuses on optimizing the YOLOv7-based method for PCBA defect detection. In this method, the Mish, a smoother function, replaces the Leaky ReLU activation function of YOLOv7, effectively expanding the network's information processing capabilities. Concurrently, a Squeeze-and-Excitation attention mechanism (SEAM) has been integrated into the head of the model, significantly augmenting the precision of small target defect detection. Additionally, considering angular loss, compared to the CIoU loss function in YOLOv7, the SIoU loss function in the paper enhances robustness and training speed and optimizes inference accuracy. In terms of data preprocessing, this study has devised a brightness adjustment data enhancement technique based on split-filtering to enrich the dataset while minimizing the impact of noise and lighting on images. The experimental results under identical training conditions demonstrate that our model exhibits a 9.9% increase in mAP value and an FPS increase to 164 compared to the YOLOv7. These indicate that the method proposed has a superior performance in PCBA defect detection and has a specific application value.