首页> 外文会议>International Conference on Computer Vision, Image and Deep Learning >Multiple Target Tracking Based on Position Matching and Feature Correlation
【24h】

Multiple Target Tracking Based on Position Matching and Feature Correlation

机译:基于位置匹配和特征相关的多目标跟踪

获取原文

摘要

Aiming at real-time application of multiple target tracking in automatic driving, an adaptive online multiple target tracking method based on maximum matching of IoU is proposed, which solves the mismatch problem of fixed IoU threshold matching. In order to solve the problem of trajectory fracture caused by occlusion, an offline method to extract fHOG feature and deep feature is raised to determine whether a connection is required after online multiple target tracking is completed, and the validity of this method is verified by experiments.
机译:旨在在自动驾驶中实时应用多个目标跟踪,提出了一种基于IOU最大匹配的自适应在线多目标跟踪方法,其解决了固定IOU阈值匹配的不匹配问题。为了解决由遮挡引起的轨迹断裂的问题,提取了提取FHOG特征和深度特征的离线方法,以确定在在线多目标跟踪完成后是否需要连接,并且通过实验验证该方法的有效性。

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