도로교통량 조사를 위한 12종 차종 분류 방법
Vehicle Type Classification Method for Road Traffic Surveys
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This paper proposes a novel method for effectively classifying 12 vehicle types required for road traffic surveys by utilizing deep learning techniques. In particular, it focuses on the trailer vehicle types, classified as types 8 to 12, which have been challenging in previous research due to data scarcity. A zero-shot learning approach, Grounding DINO, is employed to extract key features that can distinguish these trailer types, addressing the data imbalance issue. This method enables accurate classification of the underrepresented vehicle types, leading to efficient classification across all 12 types. To the best of the authors' knowledge, this is the first attempt to classify 12 vehicle types required for road traffic surveys using publicly available video data.
Abstract
This paper proposes a novel method for effectively classifying 12 vehicle types required for road traffic surveys by utilizing deep learning techniques. In particular, it focuses on the trailer vehicle types, classified as types 8 to 12, which have been challenging in previous research due to data scarcity. A zero-shot learning approach, Grounding DINO, is employed to extract key features that can distinguish these trailer types, addressing the data imbalance issue. This method enables accurate classification of the underrepresented vehicle types, leading to efficient classification across all 12 types. To the best of the authors' knowledge, this is the first attempt to classify 12 vehicle types required for road traffic surveys using publicly available video data.
참고문헌 (14)
- J. H. Park, "A Study on the Performance Improvement of Vehicle Types Classification", Proceedings of the Korean Institute of Communications and Information Sciences Conference, pp.967~968 (2021)
- J. M. Johnson, "Survey on Deep Learning with Class Imbalance", Journal of Big Data, 6(1), pp.1~54 (2019)
- Y. T. Sohn, "A Study on the Improvement of Vehicle Classification Criteria for Road Traffic Volume Survey", Journal of the Korean Society of Transportation, 19(3), pp.153~165 (2001)
- J. S. Oh, "Vehicle Classification Plan for Two-Axle Vehicles Using Vehicle Shape Data", Journal of Korean Intelligent Transport Systems Society, 10(5), pp.47~52 (2011)
- J. M. Kim, "A Study on Road Traffic Volume Survey Using Vehicle Specification DB", J. Korea Inst. Intell. Transp. Syst, 22(2), pp.93~104 (2023)
- J. Redmon, "You only Look Once: Unified, Real-time Object Detection", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.779~788 (2016)
- J. Redmon, "YOLO9000: Better, Faster, Stronger", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.7263~7271 (2017)
- M. S. Kang, "Method for Obtaining Better Traffic Survey Data", Electronics, 10(7), pp.833~ (2021)
- P. K. Kim, "Vehicle Type Classification Using Bagging and Convolutional Neural Network on Multi View Surveillance Image", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.41~46 (2017)
- Road Traffic Volume Survey Guidelines, "Ministry of Land, Infrastructure and Transport", Ministry of Land, Infrastructure and Transport, ~, pp.Ministry of Land, Infrastructure and Transport Decree No. 362 (2023)