田爱军,蔡旭阳,陈 玮,罗 伟,尹彦卿.无人机端路面车辆违停检测及取证系统[J].测控技术,2021,40(5):67-74
无人机端路面车辆违停检测及取证系统
Vehicle Illegal Parking Detection and Evidence Collection System on UAV
  
DOI:10.19708/j.ckjs.2021.05.012
中文关键词:  违章停车检测  深度学习  目标检测  无人机  红外测温
英文关键词:illegal parking detection  deep learning  object detection  UAV  infrared temperature
基金项目:
作者单位
田爱军 航空工业金城集团有限公司 
蔡旭阳 航空工业金城集团有限公司 
陈 玮 航空工业金城集团有限公司 
罗 伟 航空工业金城集团有限公司 
尹彦卿 航空工业金城集团有限公司 
摘要点击次数: 33
全文下载次数: 20
中文摘要:
      路边车辆违停现象日益严重,而针对路边违停现象的检测主要依赖于交警巡逻和固定相机抓拍,其时效性和灵活性不足。提出一种基于深度学习技术的无人机端路面车辆违停检测及取证系统,包含无人机车辆违停巡检和车辆违停取证两种模式。利用无人机高效灵活的优势,结合深度学习车辆检测技术检测路边车辆违停现象,并对深度学习目标检测算法进行了优化,得出更加贴合的检测框,同时辅以红外测温技术,进一步降低车辆违停的误判情况。选取了多个不同场景,分多天时间进行了多次实验。通过实验证明,本系统运行速度快、检测精度高、检测区域灵活,能保障道路巡检的及时性,减轻交管部门工作压力,为居民的交通便利及交通安全提供保障。
英文摘要:
      The phenomenon of illegal parking is becoming more and more serious,and the detection of illegal parking mainly depends on traffic police patrol and fixed camera capture,which is lack of timeliness and flexibility.A vehicle illegal parking detection and evidence collection system on UAV is proposed,which includes two modes:vehicle illegal parking inspection and vehicle illegal parking forensics.Using the advantages of UAVs high efficiency and flexibility,combined with deep learning vehicle detection technology to detect the phenomenon of roadside vehicles illegal parking,the deep learning object detection algorithm is optimized,and a more fitting detection bounding box is obtained.At the same time,infrared temperature measurement technology is supplemented to further reduce the misjudgment of vehicles illegal parking.A number of different scenes are selected and experiments are carried out in several days.The experiments show that the system has the advantages of fast running speed,high detection accuracy and flexible detection area,which can ensure the timeliness of road inspection,and reduce the work pressure of vehicle management,and provide protection for the residents traffic convenience and safety.
查看全文  查看/发表评论  下载PDF阅读器
关闭