카메라를 활용한 딥러닝 기반 자율주행자동차의 사각 지대 객체 검출 및 경고 시스템에 관한 연구

다국어 초록 (Multilingual Abstract)

In this paper, we propose blind spot area detection and warning system for autonomous driving automobiles based on deep running using a camera to prevent accidents caused by autonomous vehicles due to angle limitation of side mirrors. Recently, algorithms capable of real-time object detection through deep learning have attracted attention in the image processing field of autonomous driving. In this paper, SSD(Single Shot Multibox Detector) algorithm is used for detection of blind spot objects. SSD has an accuracy similar to that of conventional Faster R-CNN and has the advantage of detecting objects faster than YOLO(You Only Look Once). Experimental results show that we can solve the trade-off problem of accuracy and real-time performance of existing algorithm by applying SSD in well-balanced manner, and confirm warning and real-time detection possibility of blind spot objects of autonomous vehicles.

목차 (Table of Contents)

• Abstract
• 1. 서론
• 2. 관련 연구
• 3. 제안하는 연구방법
• 4. 실험 결과
• 5. 결론
• References