ETRI-Knowledge Sharing Plaform

ENGLISH

성과물

논문 검색
구분 SCI
연도 ~ 키워드

상세정보

학술지 Method for Obtaining Better Traffic Survey Data
Cited 0 time in scopus Download 11 time Share share facebook twitter linkedin kakaostory
저자
강미선, 김병근, 임길택, 조유제
발행일
202104
출처
Electronics, v.10 no.7, pp.1-16
ISSN
2079-9292
출판사
MDPI
DOI
https://dx.doi.org/10.3390/electronics10070833
협약과제
21ZD1100, 대경권 지역산업 기반 ICT 융합기술 고도화 지원사업, 문기영
초록
Road traffic surveys determine the number and type of vehicles passing by a specific point over a certain period of time. The manual estimation of the number and type of vehicles from images captured by a camera is the most commonly used method. However, this method has the disadvantage of requiring high amounts of manpower and cost. Recently, methods of automating traffic volume surveys using sensors or deep learning have been widely attempted, but there is the disadvantage that a person must finally manually verify the data in order to ensure that they are reliable. In order to address these shortcomings, we propose a method for efficiently conducting road traffic volume surveys and obtaining highly reliable data. The proposed method detects vehicles on the road from CCTV (Closed-circuit television) images and classifies vehicle types using deep learning or a similar method. After that, it automatically informs the user of candidates with a high probability of error and provides a method for efficient verification. The performance of the proposed method was tested using a data set collected by an actual road traffic survey company. As a result, we proved that our method shows better accuracy than the previous method. The proposed method can reduce the labor and cost in road traffic volume surveys, and increase the reliability of the data due to more accurate results.
키워드
Deep learning, Vehicle classification, Vehicle count, Verification method
KSP 제안 키워드
Closed Circuit Television(CCTV), Data sets, Probability of Error, Road traffic, Similar method, Traffic volume, Vehicle classification, Vehicle type, deep learning(DL), highly reliable, reliable data
본 저작물은 크리에이티브 커먼즈 저작자 표시 (CC BY) 조건에 따라 이용할 수 있습니다.
저작자 표시 (CC BY)