Earticle

초록

영어

As a result of analyzing domestic traffic accidents, the number of fatalities caused by black ice-related accidents is as much as 3.7 than that of snowy road accidents. Unlike ordinary snow, black ice is difficult to detect with the naked eye, and it is hard for drivers to distinguish it within a second while driving. Therefore, the plan is to apply computer vision to detect black ice in real-time. To accomplish this task, three Artificial Intelligence models were compared and analyzed. The Artificial IntelligenceI models used are CNN, YOLOv5, and YOLOv8. The CNN model detected 258 out of 285 images where black ice occurred, with an accuracy of 87%. YOLOv5 detected 270 out of 285 images, achieving an accuracy of 95%. Lastly, YOLOv8 showed the highest accuracy, detecting 280 out of 285 images with an accuracy of 98%. In the future, this research will be combined with real-time CCTV installed by the National Traffic Information Center to detect black ice in real-time during winter and provide solutions to reduce the likelihood of traffic accidents caused by black ice.

목차

Abstract
1. INTRODUCTION
2. ARTIFICIAL INTELLIGENCE MODEL DESIGN
2.1 System Structure
2.2 Dataset
2.3 CNN model design
2.4 YOLO model design
3. IMPLEMENTATION
3.1 Development environment
3.2 Result
4. CONCLUSIONS
REFERENCES

키워드

저자

• Eun-Chong Park [ Department of Computer Engineering, Honam University, Korea ]
• Kyu-Ha Kim [ Department of Computer Engineering, Honam University, Korea ]
• Sang-Hyun Lee [ Department of Computer Engineering, Honam University, Korea ] Corresponding Author

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  International Journal of Advanced Culture Technology(IJACT)

• 간기

  계간

• pISSN

  2288-7202

• eISSN

  2288-7318

• 수록기간

  2013~2025

• 등재여부

  KCI 등재

• 십진분류

  KDC 600 DDC 700