• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.651-654
• /
• 2021

Face anti-spoofing (FAS) techniques play a significant role in the defense of facial recognition systems against spoofing attacks. Existing FAS methods achieve the great performance depending on annotated additional modalities. However, labeling these high-cost modalities need a lot of manpower, device resources and time. In this work, we proposed to use self-transforming modalities instead the annotated modalities. Three different modalities based on frequency domain and temporal domain are applied and analyzed. Intuitive visualization analysis shows the advantages of each modality. Comprehensive experiments in both the CNN-based and transformer-based architecture with various modalities combination demonstrate that self-transforming modalities improve the vanilla network a lot. The codes are available at https://github.com/chenmou0410/FAS-Challenge2021.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.52-54
• /
• 2022

Aerial images are being one of the important aspects of satellite imagery, delivers effective information on landcovers. Their special characteristics includes the viewpoint from space which clarifies data related to land examining processes. Aerial images taken by satellites employed radio waves to wirelessly transmit images to ground stations. Due to transmission errors, images get distorted and unable to perform in landcover examining. This paper proposes an aerial image enhancement method using deep autoencoders. A properly trained autoencoder can enhance an aerial image to a considerable level of improvement. Results showed that the achieved enhancement is better than that was obtained from traditional image denoising methods.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.295-296
• /
• 2022

작물을 경작 해충과 질병은 오랫동안 주요 관심사였다. 농업에서 병해충을 탐지하기 위해 전통적인 방법을 사용하는 것은 더 이상 높은 효율성을 제공하지 않는다. 오늘날 과학과 인공 지능의 폭발적인 발달로 인해 농업분야의 연구원들은 병해충을 탐지하기 위해 딥 러닝을 적용하고 있다. 최근에 다양한 분야의 문제들을 해결하기 위해 수많은 모델들이 발표되었지만, 많은 병해충 진단 딥러닝을 사용한 방법들은 하드웨어 리소스를 낭비하고 실제 농장에서 사용하기 어렵다. 따라서 본 논문에서는 작물의 병해충을 분류하기 위해 Select Kernel Attention(SK Attention)을 Channel Attention Module 로 변경하여 Decoupling-and-Attention network (DANet)을 하드웨어 리소스 사용을 최소화한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.594-596
• /
• 2022

Over the last decades, deep neural networks have demonstrated significant success in various tasks. To address the special vision task, choosing a hot network as backbone to extract feature is a common way in both research and industry project. However, the choice of backbone usually requires the expert experience and affects the performance of the classification task. In this work, we propose a novel idea to support backbone decision-making by exploring the feature attribution and weights distribution of hidden layers from various backbones. We first analyze the visualization of feature maps on different size object and different depth layers to observe learning ability. Then, we compared the variance of weights and feature in last three layers. Based on analysis of the feature and wights, we summarize the traits and commonalities of existing networks.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.653-655
• /
• 2022

Face aging aims to synthesize future face images by reflecting the age factor on given faces. In recent years, deep learning-based approaches have made outstanding progress in simulating the aging process of the human face. However, generating accurate and high-quality aging faces is still intrinsically difficult. We propose a new method that incorporates gender information into the model, which achieves comparable and stable performance. Experimental results demonstrate that our method can preserve the identity well and generate diverse aged faces.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.300-302
• /
• 2022

Palm vein identification has attracted attention due to its distinct characteristics and excellent recognition accuracy. However, many contactless palm vein identification systems suffer from the issue of having low-quality palm images, resulting in degradation of recognition accuracy. This paper proposes the use of U-Net architecture to correctly segment the vascular blood vessel from palm images. Attention gate mechanism and residual block are also utilized to effectively learn the crucial features of a specific segmentation task. The experiments were conducted on CASIA dataset. Hessian-based Jerman filtering method is applied to label the palm vein patterns from the original images, then the network is trained to segment the palm vein features from the background noise. The proposed method has obtained 96.24 IoU coefficient and 98.09 dice coefficient.

• Journal of the Korean Institute of Traditional Landscape Architecture
• /
• v.38 no.1
• /
• pp.1-9
• /
• 2020

This paper investigates spatial composition and landscape characteristics of the royal tomb(Jangneung in Paju) where the 12th King(Injo) and the Queen(InnYeol) of the Joseon Dynasty are buried. Based on the investigations, the study suggests the management plan of the Royal tomb at the end. The study includes research on ancient literature and the on-site survey to estimate the ancient landscape architecture of the Royal tomb. Thereby, it is intended to provide the basic data for the identification, preservation, and restoration of the landscape architecture. As a result of the study, first, Jangneung is confirmed that it is a typical formal landscape structure of the Royal tomb in the Joseon Dynasty. Second, Ecological resources around the royal tombs are consistent with previous records and current field research. Third, although the Japanese colonial rule damaged much of the Royal tomb, it preserves much of its core facilities. However, the restoration of ancillary facilities is required, and a World Heritage Conservation Management Plan is required that complies with the ICOMOS Charter.

• Journal of Veterinary Clinics
• /
• v.24 no.4
• /
• pp.584-587
• /
• 2007

The purpose of this study was to estimate the standard of normal spleen size and its variations in relation to age, body weight(BW) and 1st lumbar vertebra($L_1$) length using radiographs. The Maximum transverse length (MTL) and maximum vertical length(MVL) of the spleen were obtained from ventrodorsal abdominal radiographs of 134 dogs which underwent radiography for indications unrelated to splenic disease. Splenic size did not significantly correlated with the age. MTL and MVL increased with BW according to the nonlinear correlation. Also, MTL and MVL increased with $L_1$ length according to the linear correlation.

• Korean Journal of Veterinary Research
• /
• v.31 no.2
• /
• pp.217-222
• /
• 1991

Most progesterone enzyme immunoassays(EIA) are used liquid phase double-antibody separation. These methods consume considerable time and reagents because of the requirements for several washing and centrifugation steps involving the reactants. Because of there several problems, we were prompted to develop an effective EIA system by the use of higher titer of progesterone antiserum free of anti-bovine serum albumin antibodies (anti-BSA). The results obtained were as follows. 1. The antibody of progesterone antiserum was high as $1.5{\times}10^5$. 2. Percent activity bound of progesterone antiserum was about 77 at a dilution to $5{\times}10^3$ times. 3. Progesterone antiserum was contained a large amount of anti-BSA antibodies. 4. The anti-BSA was completely absorbed by using of polymerised BSA. 5. The molecular weight of albumin polymer (polymerised BSA) obtained by using 2.5% glut. araldehyde was $5{\times}10^5$.

• The KIPS Transactions:PartB
• /
• v.18B no.6
• /
• pp.385-388
• /
• 2011

Patients' privacy protection is a heated issue in medical business, as medical information in digital format transmit everywhere through networks without any limitation. A current protection method for brain images is to deface from the brain image for patient's privacy. However, the defacing process often removes important brain voxels so that the defaced brain image is damaged for medical analysis. An ad-hoc method is proposed to conceal patient's identification by adding cylindrical mask, while the brain keep all important brain voxels. The proposed lossless deformation of brain image is verified not to loose any important voxels. Futhermore, the masked brain image is proved not to be recognized by others.