• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International conference on neural information processing >Pedestrian Counting System Based on Multiple Object Detection and Tracking
【24h】

Pedestrian Counting System Based on Multiple Object Detection and Tracking

机译：基于多目标检测与跟踪的行人计数系统

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

With the increasing demands on video surveillance and business promotion, effective pedestrian counting in surveillance environments has become a hot research topic in computer vision. In this paper, we implement a pedestrian counting system based on multiple object detection and tracking. Region proposal network (RPN) and Real Adaboost classifier are employed to train a head-shoulder detector with high accuracy. We utilize the DSST algorithm to track the position transformations and the size changes of pedestrians. By combining human detection with object tracking together and using detection results to optimize the tracking algorithm, the pedestrian counting system is developed with high robustness against occlusions. We evaluated the system on the videos recorded in the subway station. The results showed that our system achieves a high accuracy and can be used for pedestrian counting in crowded public places.
机译：随着对视频监控和业务推广的需求日益增长，在监控环境中有效的行人计数已成为计算机视觉中的热门研究主题。在本文中，我们实现了基于多目标检测和跟踪的行人计数系统。区域提议网络（RPN）和Real Adaboost分类器用于训练高精度的头肩检测器。我们利用DSST算法来跟踪行人的位置变换和大小变化。通过将人类检测与对象跟踪结合在一起，并使用检测结果优化跟踪算法，开发出了具有高度鲁棒性的遮挡行人计数系统。我们根据地铁站录制的视频对系统进行了评估。结果表明，我们的系统具有很高的精度，可用于拥挤的公共场所的行人计数。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号