Real-Time Lane-Vehicle Detection and Tracking System

CORC > 自动化研究所 > 中国科学院自动化研究所 > 精密感知与控制研究中心 > 精密感知与控制

	Real-Time Lane-Vehicle Detection and Tracking System
	Huang G(黄冠)1 ; Wang Xingang1 ; Wu Wenqi1 ; Zhou Han 2; Wu Yuanyuan 2; Huang G(黄冠)
	2016-05
会议日期	2016-5-28
会议地点	中国银川
关键词	Lane Detection And Tracking Vehicle Detection Vehicle Tracking Convolutional Neural Networks
英文摘要	Leading vehicle localization is the key part of intelligent driving system. This paper presents a novel and effective lane-vehicle detection and tracking system to achieve real-time and robust leading vehicle localization. The system is composed of (i) lane detection and tracking, (ii) sky and road region recognition, (iii) vehicle detection and tracking. First, lane detection and sky and road recognition are performed to reduce the possible region of leading vehicle. Then vehicle detection is applied to get all vehicles in per image. Combined with ROI information, leading vehicle is accurately localized. Finally, vehicle tracking is used to make the whole system more fast and robust. The core part of the whole system is vehicle detection. A novel and effective method based on convolutional neural networks is used to achieve high-quality and real-time vehicle detection. The experimental results validate the effectiveness and robustness of our algorithm. In addition, our system runs at real-time of 10 frames per second.
会议录	Chinese Control and Decision Conference (CCDC)
内容类型	会议论文
源URL	[http://ir.ia.ac.cn/handle/173211/11790]
专题	精密感知与控制研究中心_精密感知与控制
通讯作者	Huang G(黄冠)
作者单位	1.中国科学院自动化研究所 2.北京信息科技大学
推荐引用方式 GB/T 7714	Huang G,Wang Xingang,Wu Wenqi,et al. Real-Time Lane-Vehicle Detection and Tracking System[C]. 见:. 中国银川. 2016-5-28.