• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.15-17
• /
• 2023

ViT(Vision Transformer)는 트랜스포머 구조에 이미지를 패치들로 나눠 한꺼번에 인풋으로 입력하는 모델이다. CNN 기반 모델보다 더 적은 훈련 계산량으로 다양한 이미지 인식 작업에서 SOTA(State-of-the-art) 성능을 보이면서 다양한 비전 작업에 ViT 를 적용하는 연구가 활발히 진행되고 있다. 하지만, ViT 모델도 AI 모델 훈련시에 생성된 그래디언트(Gradients)를 이용해 원래 사용된 훈련 데이터를 복원할 수 있는 모델 역전 공격(Model Inversion Attacks)에 안전하지 않음이 증명되고 있다. CNN 기반의 모델 역전 공격 및 방어 기법들은 많이 연구되어 왔지만, ViT 에 대한 관련 연구들은 이제 시작 단계이고, CNN 기반의 모델과 다른 특성이 있기에 공격 및 방어 기법도 새롭게 연구될 필요가 있다. 따라서, 본 연구는 ViT 모델에 특화된 모델 역전 공격 및 방어 기법들의 특징을 서술한다.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.2
• /
• pp.83-90
• /
• 2023

Blurred image is an important factor in lowering image recognition rates in Computer vision. This mainly occurs when the camera is unstablely out of focus or the object in the scene moves quickly during the exposure time. Blurred images greatly degrade visual quality, weakening visibility, and this phenomenon occurs frequently despite the continuous development digital camera technology. In this paper, it replace the modified building module based on the Deep multi-patch neural network designed with convolution neural networks to capture details of input images and Attention techniques to focus on objects in blurred images in many ways and strengthen the image. It measures and assigns each weight at different scales to differentiate the blurring of change and restores from rough to fine levels of the image to adjust both global and local region sequentially. Through this method, it show excellent results that recover degraded image quality, extract efficient object detection and features, and complement color constancy.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.3
• /
• pp.140-147
• /
• 2024

In the Republic of Korea, people with hearing impairments are the second-largest demographic within the registered disability community, following those with physical disabilities. Despite this demographic significance, research on sign language translation technology is limited due to several reasons including the limited market size and the lack of adequately annotated datasets. Despite the difficulties, a few researchers continue to improve the performacne of sign language translation technologies by employing the recent advance of deep learning, for example, the transformer architecture, as the transformer-based models have demonstrated noteworthy performance in tasks such as action recognition and video classification. This study focuses on enhancing the recognition performance of sign language translation by combining transformers with 3D-CNN. Through experimental evaluations using the PHOENIX-Wether-2014T dataset [1], we show that the proposed model exhibits comparable performance to existing models in terms of Floating Point Operations Per Second (FLOPs).

• Journal of the Korean Society for information Management
• /
• v.22 no.4 s.58
• /
• pp.79-95
• /
• 2005

The alms of this study is to analyse the user's behavior, satisfaction, difficulties and selection of retrieval keywords for the use of Web-based OPAC in the College students. The methods of the questionnaire and the interview was applied to get the data and processed by using SPSSWIN 10.1. Several research results was proved the hypothesis such as differences between major subject of students in their fields. Furthermore, based on the result of this analysis, another purpose is to come up with the Improvements of functions prompting difficulties and answers to problems found in the Web OPAC, helping them to use the Web OPAC efficiently.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.1
• /
• pp.29-34
• /
• 2022

We propose a novel network architecture and build dataset for recognizing clickable objects on mobile device screens. The data was collected based on clickable objects on the mobile device screen that have numerous resolution, and a total of 24,937 annotation data were subdivided into seven categories: text, edit text, image, button, region, status bar, and navigation bar. We use the Deconvolution Single Shot Detector as a baseline, the backbone network with Squeeze-and-Excitation blocks, the Single Shot Detector layer structure to derive inference results and the Feature pyramid networks structure. Also we efficiently extract features by changing the input resolution of the existing 1:1 ratio of the network to a 1:2 ratio similar to the mobile device screen. As a result of experimenting with the dataset we have built, the mean average precision was improved by up to 101% compared to baseline.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.7
• /
• pp.279-286
• /
• 2021

Recently, We are carrying out a policy of physical distancing of at least 1m from each other to prevent the spreading of COVID-19 disease in public places. In this paper, we propose a method for measuring distances between people in real time and an automation system that recognizes objects that are within 1 meter of each other from stereo images acquired by drones or CCTVs according to the estimated distance. A problem with existing methods used to estimate distances between multiple objects is that they do not obtain three-dimensional information of objects using only one CCTV. his is because three-dimensional information is necessary to measure distances between people when they are right next to each other or overlap in two dimensional image. Furthermore, they use only the Bounding Box information to obtain the exact coordinates of human existence. Therefore, in this paper, to obtain the exact two-dimensional coordinate value in which a person exists, we extract a person's key point to detect the location, convert it to a three-dimensional coordinate value using Stereo Vision and Camera Calibration, and estimate the Euclidean distance between people. As a result of performing an experiment for estimating the accuracy of 3D coordinates and the distance between objects (persons), the average error within 0.098m was shown in the estimation of the distance between multiple people within 1m.

• Journal of Digital Convergence
• /
• v.12 no.4
• /
• pp.311-318
• /
• 2014

This paper is to introduce a lighting source controller using high brightness LED to create a clear and reliable condition for an accurate measurement and testing, which is a core technology in clinical image system and mechanical automation system. This controller is designed to supply a stable power in a constant-current system by installing a high brightness LED driver, and to improve the reproducibility of brightness by using 32-bit ARM processor core, dividing brightness quantity into 256 levels, making the remote control and the external interface possible, and preventing and digitizing the brightness inaccuracy caused by errors of resistance values. This controller enables the lighting range to be wide and possible in a low lighting level compared to analog, adds the RS-485 communication function, and makes it for the users to control the on-off function and the dimming level by receiving date from an external device.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.45 no.6
• /
• pp.149-156
• /
• 2008

SLAM(Simultaneous Localization and Mapping) system is to find a global position and build a map with sensing data when an unmanned-robot navigates an unknown environment. Two kinds of system were developed. One is used distance measurement sensors such as an ultra sonic and a laser sensor. The other is used stereo vision system. The distance measurement SLAM with sensors has low computing time and low cost, but precision of system can be somewhat worse by measurement error or non-linearity of the sensor In contrast, stereo vision system can accurately measure the 3D space area, but it needs high-end system for complex calculation and it is an expensive tool. In this paper, we implement the SLAM system using a single camera image and a PSD sensors. It detects obstacles from the front PSD sensor and then perceive size and feature of the obstacles by image processing. The probability SLAM was implemented using the data of sensor and image and we verify the performance of the system by real experiment.

• Journal of the Korea Convergence Society
• /
• v.8 no.11
• /
• pp.15-20
• /
• 2017

Recently, with the development of the IT industry, interest in computer convergence technology is increasing in various fields. Especially, in the semiconductor field, a vision system that uses a camera and computer convergence is often used to inspect semiconductor device defects in the production process. Various systems have been studied to remove noise, which is a major cause of degradation in processing of data related to these image processing systems. In this paper, we try to detect defects in BGA (Ball Grid Array) package devices by recognizing defects in advance during mass production. We propose a measurement system using a Gaussian filter, a Median filter, and an Average filter, which are widely used for noise reduction of image data Applying the proposed system to the manufacturing process of the BGA package can be used to judge whether the defect is good or not, and it is expected that productivity will be improved.

• Journal of Bio-Environment Control
• /
• v.12 no.1
• /
• pp.30-37
• /
• 2003

An expert system was developed for the stress diagnosis of cucumber and tomato using FoxPro. The principle points in building the system were integration with Korean, effective processing of mass information, and easy access for non-experts such as farmers. The method of inferencing was forward chaining based on pattern matching. Knowledge base was expressed with IF∼THEN rules and was expressed in the form of tree. Also, the expert system was designed so that additions and modifications of all information could easily be performed on windows. The results tested by farmers with the developed system showed that the expert system was reliable for the practical use. It was expected the expert system could be directly applied to the stress diagnosis of other vegetable plants by modifying only data bases.