http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
임베디드 시스템을 위한 두 실시간 객체 탐지 알고리즘의 성능 비교
염수형(Soo-Hyung Yeom),강준영(Junyeong Kang),류지수(Ji-Su Ryu),최병조(Byoung-Jo Choi),이명렬(Myeong-Lyeol Lee) 대한전자공학회 2022 대한전자공학회 학술대회 Vol.2022 No.11
In our experiment, we employed two different object detection schemes for embedded systems, namely, YOLOv3-tiny and FOMO, in the Raspberry Pi 4B environment to detect honeybees at the gate of beehives in real time. We compared the two schemes in terms of their accuracies, operating temperatures, and processing times. Regarding their accuracies, YOLOv3-tiny displayed 96.2%, while FOMO resulted in 92.7%. However, FOMO required a processing time of 103ms per frame, while 1,235ms for YOLOv3-tiny, resulting in 12times faster processing speed for FOMO.
e-Learning 국제교류협력 프로젝트를 통한 초등학생들의 영어학습 사례연구
염수경(Yeom, Sukyung),정행(Jung, Haeng) 팬코리아영어교육학회(구 영남영어교육학회) 2009 영어교육연구 Vol.21 No.3
This paper aims not only to demonstrate how to implement e-Learning International Exchange Cooperation Project in elementary school English classes but also to investigate the effects of the project on the elementary school students' English capability and attitudes. This is a case study of a mixed group of 11 students (five 4th graders, one 5th grader, five 6th graders) who participated in performing the projects and exchanged the outputs with the overseas students by e-mails or internet blogs for 8 weeks (16 lessons). For this study, three research questions were posed. 1) Does the project improve the students' English reading ability? 2) Does the project improve the students' English writing ability? 3) Does the project motivate the students' interests and confidence in English? To answer the research questions 1) and 2), each student's performance and outputs for the projects were collected and analyzed in terms of 3 levels (excellent, good, need effort.) To answer the research question 3), the teacher's observation journal for each student's attitude was examined. The results indicated that the projects influenced the students' English learning and they help to develop the students' reading and writing abilities. It was also found that the projects affected the students' attitudes and led to the increase of the students' interests and confidence in English.
CCW 알고리즘 기반 꿀벌 카운팅 시스템 및 꿀벌 분류 시스템
류지수(Ji-Su Ryu),염수형(Soo-Hyung Yeom),강준영(Junyeong Kang),최병조(Byoung-Jo Choi),이명렬(Myeong-Lyeol Lee) 한국정보기술학회 2022 Proceedings of KIIT Conference Vol.2022 No.6
대표적인 화분매개곤충인 꿀벌의 경제적 가치는 전 세계적으로 농작물 화수분 면에서 매우 크다. 하지만 농약, 기후변화 등 환경문제로 꿀벌의 개체 수가 감소하고 있다. 따라서 꿀벌의 감소량을 파악한 후 더 큰 감소로 이어지는 사태를 조기에 방지하는 것이 중요하다. 본 논문에서는 선분 교차 판별 알고리즘인 CCW(Counter-ClockWise)를 이용해 출입하는 꿀벌의 수를 자동으로 계산해주는 딥 러닝 기반 꿀벌 카운팅 시스템을 구현하였다. 또한, 벌의 종류에 따른 개체 수 자동 집계를 위해 직접 촬영한 여왕벌, 일벌, 수벌의 이미지를 사용하여 개체별 학습을 진행하였다. 총 600장의 이미지 데이터를 이용하여 다중객체 인식 면에서 0.03%의 loss rate를 얻을 수 있었다. 본 연구결과를 활용하면 분봉을 예측하여 꿀벌 개체 수 감소를 최소화하는 시스템을 구축할 수 있을 것으로 기대된다. The economic value of honeybees, a representative pollinating insect, is very large in terms of crop pollination worldwide. However, the number of honeybees is decreasing due to environmental problems such as pesticides and climate change. Therefore, it is important to identify the amount of decrease in honeybees and prevent the situation leading to a greater decrease in early stage. In this paper, we implement a deep learning-based honeybee counting system that automatically calculates the number of honeybees entering and leaving employing CCW (Counter-ClockWise), a line segment cross-determination algorithm. Our deep learning process utilized the image data of queen bees, worker bees, and drones for classification of the bees. Using 600 images, it was possible to obtain a loss rate 0.03% accuracy in terms of multi-class recognition. It is expected that our automatic honeybee counting system may be used to reduce the colony collapse disorder by giving an early warning to the beekeepers.
강준영(Junyeong Kang),류지수(Ji-Su Ryu),염수형(Soo-Hyung Yeom),최병조(Byoung-Jo Choi),이명렬(Myeong-Lyeol Lee) 한국정보기술학회 2022 Proceedings of KIIT Conference Vol.2022 No.12
꿀벌의 수분활동은 생태계 유지에 중요한 영향을 끼친다. 하지만 오늘날 꿀벌 개체 수는 전 세계적으로 감소하는 추세이다. 본 연구에서는 꿀벌 개체 수 모니터링을 위해 양봉 환경에서 꿀벌 활동 이미지를 전송하고 해당 이미지에 대한 객체 탐지를 진행하였다. 탐지 결과를 데이터베이스에 저장하고 이미지와 탐지 결과를 웹페이지에 시각화하는 통합 시스템을 구현하였다. 양봉 환경에서 이미지 데이터를 수집하는 과정에서 저전력 소형 장치인 ESP-EYE를 사용하였으며, 객체 탐지는 YOLOv5 모델에 1752개의 이미지 데이터를 학습하여 적용하였으며, 평균 객체 검출 정확도는 98%였다. 구현된 통합시스템은 향후 꿀벌 행동 특성 분석 연구에 활용될 것으로 기대된다. Honeybees pollination has an important effect on maintaining the ecosystem. Today, however, the bee population is on the global decline. In this paper, honeybee activity images at the gate of beehives are transmitted in a beekeeping environment and object detection is performed on the image. We developed an integrated system that stores detection results in a database and visualizes the images as well as the detection results on the web pages. We employed an ESP-EYE, which is a low power embedded device, for collecting image data in a beekeeping environment. Object detection is performed using YOLOv5 for 1752 image data and the mean average precision was 98%. Our integrated system may be used for analyzing behavioral characteristics of honeybees.
무균돼지 사육시설 계획을 위한 사례연구 : 공간구성 및 동선계획을 중심으로
권순정,성제경,염수청 한국의료복지시설학회 2006 의료·복지 건축 Vol.12 No.1
According to the increase of demand for human organs such as kidney, heart, pancreas, joint, and cornea for therapeutic transplantation, the production of alternative organs based on Gnotobiotic Pigs gains a lot of concerns all over the world. However, it is not common to design and build Gnotobiotic Pigs' facility, and there are only a few those facilities and planning principles for them. Considering the situation above, this paper tries to develop planning guidelines for space organization and circulation system of standardized Germ Free Pig's facility on the bases of case analysis. The results of this study are as follows. At first, four swine farms including a Gntobiotic Pig's facility has been analysed from the point of space organization and circulation system. Secondly, the space zoning of Gnotobiotic Pigs' facility has been proposed into 5 groups : pigs' area, adminstration area, operating room and laboratory, service area, and mechanical area. Space components of each group have been presented also. Finally, circulation system of Gnotobiotic Pigs' facility has been explored from a operational point of view. This, also, includes human circulation, pig's circulation, and goods' circulation. This study has some limitations because it does not consider the SOPs(standard operational policies) of that facility to the fullest measure and does not suggest space area of each part, either. Despite of some weaknesses, it is expected that this study can give some useful guidelines for the design and planning of Germ Free Pigs' facilities.