结合全局光流的孪生区域提名网络目标跟踪算法

Siamese Region Proposal Network Object Tracking Algorithm with Global Optical Flow

摘要:

基于孪生网络的跟踪器受限于孪生网络跟踪框架固有的跟踪机制和搜索区域选择机制,当目标处在被遮挡、快速运动和出视野等困难场景下时,如何稳定、鲁棒地进行目标跟踪始终是孪生网络跟踪器亟需解决的问题。为此,文章提出一种结合光流的孪生区域提名网络目标跟踪算法(GOF-SiamRPN)。通过全局光流对目标的运动趋势信息进行补充,该方法可以有效地解决在这些困难场景下的跟踪问题。在VOT2019和UAV123上的实验结果表明,相比基准方法,该算法分别取得了2.0%和1.8%的性能提升。与其他先进的跟踪器相比,该算法也取得了有竞争力的跟踪效果。

Abstract:

The Siamese network-based trackers are limited by the inherent tracking mechanism and search area selection mechanism of the Siamese network tracking framework. When the object is under challenging scenarios such as occlusion, fast motion, and out-of-view, how to perform stable and obust object tracking is always an urgent problem for Siamese trackers. To this end, in this paper, an object-tracking algorithm that combines the Siamese region proposal network with the global optical flow (GOF-SiamRPN) is proposed. By assisting the motion trend information of the object with global optical flow, the proposed method could effectively solve the potential tracking issues in these challenging scenarios. Extensive experimental results on VOT2019 and UAV123 show that the proposed method achieves a performance gain of 2.0% and 1.8% compared with the baseline method. It also achieves a competing performance compared to other state-of-the-art trackers.