VEHICLE NUMBER DETECTION AI SYSTEM FOR PEDESTRIAN SAFETY

Min-Hyuck Lim; Joon-Young Oh; Seung-Oh Jung; Jun-Moo Heo; Hyun-Joo Park

추천

검색

질문

자료유형: 학술대회자료

저자정보: Min-Hyuck Lim (Sang Myung University) Joon-Young Oh (Sang Myung University) Seung-Oh Jung (Sang Myung University) Jun-Moo Heo (Sang Myung University) Hyun-Joo Park (Sang Myung University)

저널정보: 한국엔터테인먼트산업학회 한국엔터테인먼트산업학회 학술대회 논문집 한국엔터테인먼트산업학회 2022년도 추계학술대회 논문집

발행연도: 2022.11

수록면: 63 - 67 (5page)

이용수

📌

연구주제

📖

연구배경

🔬

연구방법

🏆

연구결과

초록· 키워드

오류제보하기

With the spread of CCTV, various technologies have been developed for identification of the number of vehicles violating the law and for accurate judgment of the situation. The development of computer vision technology for processing recognized images ensures that there is no damage caused by erroneous situational judgment due to misrecognition. Recently, artificial intelligence deep learning technologies have been introduced to accurately recognize the number of vehicles that violate traffic laws using various image processing techniques. With the rapid change of society, traffic laws are also evolving at a rapid pace. We propose a deep learning algorithm for accurate situation determination of recently revised traffic laws. The proposed algorithm outputs accurate information after detecting a violation, extracting the detected vehicle license plate using image processing, and performing a number recognition process with a pre-learned cnn model. The cnn model for license plate recognition used YOLO. The YOLO learning model was used to extract areas by processing the images of vehicles, crosswalks, traffic lights, and pedestrians. In order to increase the accuracy of model validation, if the accuracy is less than 80%, it is judged that the bias is high and the image preprocessing is performed again or the loss value is reduced. Find the optimal model. In the future, it can be used as a recognition system not only in automobiles but also in various fields such as two-wheeled vehicles.

#YOLO #CNN #OCR #Deep Learning #Accuracy

참고문헌 (0)

참고문헌 신청

참고문헌이 DBpia에서 서비스 중이라면, [참고문헌 신청]을 통해 등록해보세요

함께 읽어보면 좋을 논문

논문 유사도에 따라 DBpia 가 추천하는 논문입니다. 함께 보면 좋을 연관 논문을 확인해보세요!