![](https://img-blog.csdnimg.cn/20201014180756780.png?x-oss-process=image/resize,m_fixed,h_64,w_64)
Siamese based tracker
deason_yuan
计算机科学与技术在读博士生
展开
-
Deeper and Wider Siamese Networks for Real-Time Visual Tracking--CVPR‘19 阅读
作者分析了Siamese跟踪器为什么不能很好地利用较深的网络,并设计一种新的网络架构可以使得该类跟踪器可以利用更深以及更宽的网络来提升性能。Abstract.Siamese networks have drawn great attention in visual tracking because of their balanced accuracy and speed. However, the backbone networks used in Siamese trackers are re.原创 2020-07-24 15:35:15 · 257 阅读 · 0 评论 -
SiamRPN++: Evolution of Siamese Visual Tracking with Very Deep Networks—CVPR’19 阅读
使用ResNet-50作为特征提取的网络构架,提出depth-wise和layer-wise操作针对特征进行优化,提升SiamRPN网络的性能。Abstract.Siamese network based trackers formulate tracking as convolutional feature cross-correlation between a target template and a search region.基于Siamese网络的跟踪器将跟踪任务建模为目标模板和搜索原创 2020-07-23 21:10:39 · 231 阅读 · 0 评论 -
Distractor-aware Siamese Networks for Visual Object Tracking—ECCV’18 阅读
通过主动调节具有语义样本和简单样本的数据分布,增强网络的判别能力;同时,在网络架构方面进行了深入探索,将网络宽度进行提升,增大网络容量;动态调节网络搜索范围,以适应不同尺度目标的运动。Abstract.Recently, Siamese networks have drawn great attention in visual tracking community because of their balanced accuracy and speed.最近基于Siamese的跟踪器由于精度高速原创 2020-07-23 20:29:28 · 288 阅读 · 1 评论 -
DCFNet: Discriminant Correlation Filters Network for Visual Tracking 阅读
作者将DCF视为在Siamese网络中添加的相关滤波层,并通过将网络输出定义为跟踪目标位置的概率热图来推导反向传播。将特征提取网络和相关滤波网络一起进行端到端训练。Abstract.Discriminant Correlation Filters (DCF) based methods now become a kind of dominant approach to online object tracking.基于DCF的跟踪方法已经成为一种主流的跟踪方法。The features us原创 2020-07-23 19:54:23 · 383 阅读 · 0 评论 -
High Performance Visual Tracking with Siamese Region Proposal Network—CVPR2018 阅读
作者将流行的跟踪算法分为两类,一类是基于相关滤波类并进行在线更新的跟踪算法,另一类是使用深度特征抛弃在线更新的跟踪算法,前者严重限制了跟踪速度,后者没有使用域特定信息(即某个特定的跟踪视频的信息)。针对以上不足,作者提出一个SiamRPN网络,该网络分为模板分支和检测分支。训练过程中,在相关特征图上执行proposal extraction、没有预定义好的类别信息;在跟踪过程中使用one-shot检测框架和meta-learning。Abstract.Visual object tracking h原创 2020-07-16 21:57:04 · 262 阅读 · 0 评论 -
A Twofold Siamese Network for Real-Time Object Tracking—CVPR2018 阅读
本文在SiamFC的基础上增加了语义分支,进一步提升SiamFC的判别力。Abstract.Observing that Semantic features learned in an image classification task and Appearance features learned in a similarity matching task complement each other, we build a twofold Siamese network, named SA-Sia原创 2020-07-15 21:48:25 · 369 阅读 · 0 评论 -
SINT++: Robust Visual Tracking via Adversarial Positive Instance Generation—CVPR2018 阅读
总体来说就是作者认为训练样本中正样本缺乏多样性,作者采用一个正样本生成网络来生成具有多样性的正样本用于网络的训练,并采用Siamese跟踪框架验证了算法的有效性。Abstract.Existing visual trackers are easily disturbed by occlusion, blur and large deformation. We think the performance of existing visual trackers may be limited due to原创 2020-07-14 22:33:47 · 307 阅读 · 0 评论 -
Learning Dynamic Siamese Network for Visual Object Tracking—ICCV2017 阅读
本文主要是基于SiamFC跟踪器,考虑如何使tracker能够对目标外观随时间的变化做出实时响应以及如何对背景噪声干扰进行抑制。虽然Siamese网络在跟踪器的匹配精度以及速度保持上具有巨大的潜力。但是它们在分类上仍有很大的差距。本文作者提出一种动态的Siamese网络,通过快速转换学习模型,能够有效的在线学习历史数据上的目标外观变化以及背景抑制。并给出一种基于多层特征自适应融合的方法。同时所提出的动态Siamese网络可以直接在标记的视频序列上进行训练,能够充分利用运动目标的时空信息,而不是随机选取两帧来原创 2020-07-13 21:15:42 · 338 阅读 · 0 评论 -
End-to-end representation learning for Correlation Filter based tracking 阅读
作者尝试将correlation filter (CF) 添加到深度学习网络框架中,把CF作为CNN的一个特别的层,以期待得到快速且高精度的跟踪性能。把CF在线的有效学习能力与CNN离线阶段的特征判别能力结合在一起进行端到端的训练。Abstract:The Correlation Filter is an algorithm that trains a linear template to discriminate between images and their translations. It原创 2020-07-11 15:22:27 · 450 阅读 · 0 评论 -
Fully-Convolutional Siamese Networks for Object Tracking—ECCVW2016 阅读
提出一种基于全卷积Siamese network的基本追踪算法模型.Abstract.The problem of arbitrary object tracking has traditionally been tackled by learning a model of the object’s appearance exclusively online, using as sole training data the video itself.通常任意目标的跟踪问题是通过在线学习目标的.原创 2020-07-10 14:30:10 · 224 阅读 · 0 评论 -
Siamese Instance Search for Tracking 阅读
这是一篇CVPR2016的用DL思想解决Tracking的文章。文章思想简单易懂,但是效果却很好。使用siamese深度网络,no model updating, no occlusion detection, no combination of trackers, no geometric matching,达到state-of-the-art的水平。论文旨在学习匹配机制,从大量的外部视频学...原创 2018-09-27 10:26:03 · 933 阅读 · 0 评论