计算机应用 ›› 2019, Vol. 39 ›› Issue (12): 3440-3444.DOI: 10.11772/j.issn.1001-9081.2019081427

• 第十七届中国机器学习会议(CCML 2019)论文 • 上一篇    下一篇

基于孪生检测网络的实时视频追踪算法

邓杨1,2, 谢宁1,2, 杨阳1,2   

  1. 1. 电子科技大学 计算机科学与工程学院, 成都 611731;
    2. 电子科技大学 未来媒体研究中心, 成都 611731
  • 收稿日期:2019-04-29 修回日期:2019-07-26 出版日期:2019-12-10 发布日期:2019-09-04
  • 作者简介:邓杨(1993-),男,安徽六安人,硕士研究生,主要研究方向:计算机视觉、深度学习;谢宁(1983-),男,吉林长春人,副教授,博士,CCF会员,主要研究方向:机器学习、计算机图形学;杨阳(1983-),男,辽宁大连人,教授,博士,CCF会员,主要研究方向:人工智能、多媒体信息处理。
  • 基金资助:
    国家自然科学基金资助项目(61602088);贵州省科技重大专项计划项目(20183002)。

Siamese detection network based real-time video tracking algorithm

DENG Yang1,2, XIE Ning1,2, YANG Yang1,2   

  1. 1. School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu Sichuan 611731, China;
    2. Center for Future Media, University of Electronic Science and Technology of China, Chengdu Sichuan 611731, China
  • Received:2019-04-29 Revised:2019-07-26 Online:2019-12-10 Published:2019-09-04
  • Contact: 谢宁
  • Supported by:
    This work is partially supported by the National Natural Science Foundation of China (61602088), the Major Special Plan for Science and Technology of Guizhou Province (20183002).

摘要: 目前,在视频追踪领域中,大部分基于孪生网络的追踪算法只能对物体的中心点进行定位,而在定位快速形变的物体时会出现定位不准确的问题。为此,提出基于孪生检测网络的实时视频追踪算法——SiamRFC。SiamRFC算法可直接预测被追踪物体位置,来应对快速形变的问题。首先,通过判断相似性来得到被追踪物体的中心点位置;然后,运用目标检测的思路,通过选取一系列的预选框来回归最优的位置。实验结果表明,所提SiamRFC算法在VOT2015|16|17的测试集上均有很好的表现。

关键词: 孪生网络, 物体检测, 实时视频追踪, 相似性学习, 卷积神经网络

Abstract: Currently, in the field of video tracking, the typical Siamese network based algorithms only locate the center point of target, which results in poor locating performance on fast-deformation objects. Therefore, a real-time video tracking algorithm based on Siamese detection network called Siamese-FC Region-convolutional neural network (SiamRFC) was proposed. SiamRFC can directly predict the center position of the target, thus dealing with the rapid deformation. Firstly, the position of the center point of the target was obtained by judging the similarity. Then, the idea of object detection was used to return the optimal position by selecting a series of candidate boxes. Experimental results show that SiamRFC has good performance on the VOT2015|16|17 test sets.

Key words: Siamese network, objection detection, real-time video tracking, similarity learning, Convolutional Neural Network (CNN)

中图分类号: