http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
무인기를 이용한 심층 신경망 기반 해파리 분포 인식 시스템
구정모,명현,Koo, Jungmo,Myung, Hyun 한국로봇학회 2017 로봇학회 논문지 Vol.12 No.4
In this paper, we propose a jellyfish distribution recognition and monitoring system using a UAV (unmanned aerial vehicle). The UAV was designed to satisfy the requirements for flight in ocean environment. The target jellyfish, Aurelia aurita, is recognized through convolutional neural network and its distribution is calculated. The modified deep neural network architecture has been developed to have reliable recognition accuracy and fast operation speed. Recognition speed is about 400 times faster than GoogLeNet by using a lightweight network architecture. We also introduce the method for selecting candidates to be used as inputs to the proposed network. The recognition accuracy of the jellyfish is improved by removing the probability value of the meaningless class among the probability vectors of the evaluated input image and re-evaluating it by normalization. The jellyfish distribution is calculated based on the unit jellyfish image recognized. The distribution level is defined by using the novelty concept of the distribution map buffer.
문제 공간 탐색을 통한 머신러닝 기반 오토터레인 로직 개선 개발
위경수(Kyoungsoo We),구정모(JungMo Koo),이혜연(Hye Yeon Lee),이승태(Seung Tae Lee) 한국자동차공학회 2023 한국자동차공학회 부문종합 학술대회 Vol.2023 No.5
Auto-Terrain logic was released in 2021 to enhance controllability of sport utility vehicles on every road surfaces such as paved, snowy, sandy and unpaved ways. For this, this logic uses machine learning techniques to recognize the road surface the car is driving on and automatically selects the appropriate driving mode. To improve the accuracy of the road surface recognition, post-processing has been added to the result of the learning model, and this causes the additional delay in recognizing the road surface. In this paper, to reduce the delay of the post-processing, we try to increase the accuracy of the learning model. For this, we explore the problem space which consists of various input combinations of the model. As a result, we get the improved model which increases the accuracy from 84.4% to 98.5%, and also get the efficient model which reduces the memory usage while minimizing the decrease in accuracy compared to the improved model. Also, this efficient model reduces the post-processing delay by 57% compared to the original model.
위경수(Kyoungsoo We),구정모(JungMo Koo),이혜연(Hye Yeon Lee),이승태(Seung Tae Lee) 한국자동차공학회 2023 한국자동차공학회 학술대회 및 전시회 Vol.2023 No.11
In the automotive field, identification of new failures is very important in terms of safety. However, existing techniques have limitations to identify new failures. Therefore, in this paper, we propose an unsupervised learning-based method to detect new failures that cause abnormal driving maneuver. We show that the proposed approach would be implemented with up to 2,000 times less memory while achieving higher accuracy compared to existing approaches.