• Proceedings of the Korea Information Processing Society Conference
• /
• 2017.04a
• /
• pp.976-978
• /
• 2017

포장 도로의 균열 또는 유실에 따라 발생하는 포트홀은 환경 변화에 따라 지속적으로 발생하며 이로 인한 교통사고도 지속적으로 발생한다. 포트홀 탐지를 위해 크게 3가지 방법들이 시도되고 있다. 그 중 이미지 처리를 이용한다. 포트홀은 내부에 깊이가 있으며 거친 질감을 가진다. 이러한 특성을 이용하여 포트홀을 탐지한다.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.5
• /
• pp.17-28
• /
• 2023

Potholes that occur on paved roads can have fatal consequences for vehicles traveling at high speeds and may even lead to fatalities. While manual detection of potholes using human labor is commonly used to prevent pothole-related accidents, it is economically and temporally inefficient due to the exposure of workers on the road and the difficulty in predicting potholes in certain categories. Therefore, completely preventing potholes is nearly impossible, and even preventing their formation is limited due to the influence of ground conditions closely related to road environments. Additionally, labeling work guided by experts is required for dataset construction. Thus, in this paper, we utilized the Mean Teacher technique, one of the semi-supervised learning-based knowledge distillation methods, to achieve robust performance in pothole image classification even with limited labeled data. We demonstrated this using performance metrics and GradCAM, showing that when using semi-supervised learning, 15 pre-trained CNN models achieved an average accuracy of 90.41%, with a minimum of 2% and a maximum of 9% performance difference compared to supervised learning.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.12
• /
• pp.603-611
• /
• 2020

Potholes on the road directly affect drivers' safety, satisfaction, and vehicle damage. Thus, real-time detection and response are required. Increasing frequency of patrols allows for potholes to be detected and responded to quickly, but this takes much manpower, money, and time. In addition, potholes have different occurrence characteristics depending on the rain conditions, so it is necessary to consider the optimal frequency from an economic and road-service perspective. Therefore, a quantitative analysis was done on the effects of rainfall on the occurrence characteristics of potholes. Information on the persistence, impact of rainfall intensity, and weather information was collected over a long period. Based on the results, a risk-based, optimized, and changeable road-patrol strategy is presented. The analysis results show that the probability of pothole occurrence increases by 2.4 times in rainy weather. Furthermore, the impact continues for 3 days even after the rain stops. The probability of pothole occurrence increases by 0.46% per 1 mm of rainfall, and the occurrence characteristics react sensitively to even a small amount of rain of around 1 mm. It was concluded that road patrol is required at least once every three days for an effect-free period, while twice a day is needed for the "sphere of influence" period to achieve a 95% reliability level.ys for effect-free period, while twice a day for sphere of influence period to satisfy 95% reliability level.

• Proceedings of the KAIS Fall Conference
• /
• 2009.12a
• /
• pp.526-530
• /
• 2009

본 연구은 자동차용 가솔린엔진에 장착되는 인젝터의 연료공급 특성에 대한 것으로, 가솔린 엔진의 전자제어식 포트 연료분사는 분무장치와 흡기포트의 최적화 및 분무특성이 우수해야 엔진의 성능 향상 및 배기가스 저감의 목적을 이룰 수 있다. 4홀과 12홀 인젝터의 장착각 변화와 포트 마스킹의 형상변화에 따른 벽유량을 측정?분석하였고 분무가시화 실험을 통하여 분무성장과정과 분사각, 연료미립화 및 분무도달거리를 분석하였다. 벽류측정 실험을 통하여 벽류는 미립화정도와 흡기유동과 유속에 가장 큰 영향을 받는 것으로 판단되며, 12홀 인젝터 대비 4홀 인젝터는 분무압력에 따라 분무특성의 변화량이 크게 나타났다.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.17 no.5
• /
• pp.55-64
• /
• 2018

In this paper, we propose fully convolutional neural networks based automatic detection of a pothole that directly causes driver's safety accidents and the vehicle damage. First, the training DB is collected through the camera installed in the vehicle while driving on the road, and the model is trained in the form of a semantic segmentation using the fully convolutional neural networks. In order to generate robust performance in a dark environment, we augmented the training DB according to brightness, and finally generated a total of 30,000 training images. In addition, a total of 450 evaluation DB was created to verify the performance of the proposed automatic pothole detection, and a total of four experts evaluated each image. As a result, the proposed pothole detection showed robust performance for missing.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.38 no.2
• /
• pp.319-325
• /
• 2018

Pot-holes are steadily increasing due to abnormal climate such as heavy rainfall and frequent snowfall. Pot-hole related to traffic accidents cause injuries, car damage and distress of road facilities. To reduce pot-holes, the use of an anti-stripping agent is mandatorily recommended to asphalt concrete mixture. Hydrated lime is commonly used as anti-stripping agent due to the convenience and economics. Byproduct ash from circulating fluidized bed boiler was reviewed as an anti stripping agent. According to the test results, the byproduct ash is satisfied with TSR specification using 1% to 3% by weight of the asphalt mixture. The byproduct ash was examined under various condition changes of aggregate and asphalt concrete mixture considering quality movement. According to the results, using the byproduct ash was measured average 0.87 of TSR and coverage rates of 60% after rolling bottle test. Test results also revealed that the byproduct ash showed stable performance. Using the byproduct ash to decrease pot-hole in asphalt concrete pavement is suitable for demonstrating stable performance as anti-stripping agent.

• Journal of The Geomorphological Association of Korea
• /
• v.19 no.2
• /
• pp.31-49
• /
• 2012

Satae cheon, a tributary of the Suip cheon in Yanggu, Gangwon province, is an international river extended to North Korea. Most of drainage basin area of the river was the fierce battle field during the Korean War(1950-1953) and hard to access as it located between the MDL(Military Demarcation Line) and the CCZ(Civilian Control Zone: about 10km south from MDL). By the restriction of access to the sites, most of natural landscape have been well conserved except limited use for military activities. Even the landfoms in that area were not studied, except the government's heritage reports. Satae Cheon's channel follows the Imdang fault line(N-S) to Satae-ri and flow to west to the Dutayeon area. The river meanders along geological structure or weak line at the Dutayeon area. The meandering channel was shorten by the meander cut which linked the thalweg line of meander loop ant the meander neck. As a result of this cut, the river cliff formed by the Satae cheon became the part of newly formed channel bed and the S-forms are formed. After the channel route stabilized, channel incised the rock with large potholes and undulating walls were formed. The channel width changes from 1m to 10m with restriction of the undulating walls, so this part can be regarded as inner channel or inner gorge. From the point of planar forms it also can be slot-type canyon.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.01a
• /
• pp.35-38
• /
• 2023

포트홀은 주행하는 자동차와 접촉이 이뤄지면 차체나 운전자에게 충격을 주고 제어를 잃게 하여 도로 위 안전을 위협할 수 있다. 포트홀의 검출을 위한 국내 동향으로는 진동을 이용한 방식과 신고시스템 이용한 방식과 영상 인식을 기반한 방식이 있다. 이 중 영상 인식 기반 방식은 보급이 쉽고 비용이 저렴하나, 컴퓨터 비전 알고리즘은 영상의 품질에 따라 정확도가 달라지는 문제가 있었다. 이를 보완하기 위해 영상 인식 기반의 딥러닝 모델을 사용한다. 따라서, 본 논문에서는 사전 학습된 딥러닝 모델의 정확도 향상을 위한 Feature Level Ensemble 기법을 제안한다. 제안된 기법은 사전 학습된 CNN 모델 중 Test 데이터의 정확도 기준 Top-3 모델을 선정하여 각 딥러닝 모델의 Feature Map을 Concatenate하고 이를 Fully-Connected(FC) Layer로 입력하여 구현한다. Feature Level Ensemble 기법이 적용된 딥러닝 모델은 평균 대비 3.76%의 정확도 향상을 보였으며, Top-1 모델인 ShuffleNet보다 0.94%의 정확도 향상을 보였다. 결론적으로 본 논문에서 제안된 기법은 사전 학습된 모델들을 이용하여 각 모델의 다양한 특징을 통해 기존 모델 대비 정확도의 향상을 이룰 수 있었다.

• Korean Journal of Construction Engineering and Management
• /
• v.15 no.4
• /
• pp.150-158
• /
• 2014

A bunch of asphalt roads have been damaged frequently in relation to the rapid climate change. To solve and prevent this type of problems, many nationalities in the world have performed various researches. In this regard, the objective of this study is to develop prediction model as to the number of potholes occurred in seoul. At the same time, we have utilized empirical and statistical approaches in order for us to identify factors which is affecting the actual occurrence. The predictive model was determinded by using BHS (Basic Harmony Search) algorithm. Prediction was based on the weather and traffic data as well as data occurrence data of porthole. To assess the influences which are PAR(Pitch Adjusting Rate) and HMCR(Harmony Memory Considering Rate), we determined suitability by changing the values. In the process of the determining a predictive model, the predictive model composed Training data (2011, 2012 and 2013yrs data). To determine the suitability of the model, we have utilized Testing Set (2009 and 2010 yrs data). The suitability of the basic prediction model has been from RMSE(Root Mean Squared Error), MAE(Mean Absolute Error) and Coefficient of determination.

• 한국도로학회지:도로
• /
• v.16 no.3
• /
• pp.64-69
• /
• 2014