论文阅读 | DasiamRPN

文献：DaSiamRPN: Zheng Zhu, Qiang Wang, Bo Li, Wu Wei, Junjie Yan, Weiming Hu."Distractor-aware Siamese Networks for Visual Object Tracking." ECCV (2018). [paper][github]

文章主要贡献

1.训练数据的扩充

1. 加入Detection pair (ImageNet,COCO中做数据增广)
2. negative simple in same categories (Called Distractor-aware Training)
3. negative simple in different categories (Called Distractor-aware Training)

2.Distractor Model

• 引入Distractor Model,将Proposal与exemplar的相似性度量得分减去所有之前预先得到的Distrator(NMS将网络提出的proposal去冗余,去掉classification score最高的proposal，在剩下的Distrator set 中保留score大于给定阈值的proposal)与当前proposal 的score(相似性度量)的加权和的平均

3.long term Tracking

• 当追丢时用local to global stategy 恒定step迭代的增加搜索区域的大小

具体而言

1. 扩充数据集

---

detection pairs	negative pair from same categoriess	negative pairs from different categories

2. Distractor Model

TEST:(Gamma(n)=(n-1)!quadforall ninmathbb N)
传统的SiamTracking是用求相似性度量用以下公式：

[f(x)=varphi(x)*varphi(z)+bcdotmathbf{1} ]

• 作者提出将NMS将网络提出的proposal去冗余,去掉classification score最高的proposal，在剩下的Distrator set 中保留score大于给定阈值的proposal)与当前proposal 的score(相似性度量)的加权和的平均

[q=mathop{argmax}limits_{p_{k}inmathcal{P}} f(z,p_{k})- frac{hat{alpha}-sum_{i=1}^{n}alpha_{i}f(d_{i},p_{k})} {sum_{i=1}^{n}alpha_{i}}]

(mathcal{P}) 是score在top-k的proposal, (alpha_{i})是每个干扰proposal的权重(paper中是全为1), (d_{i})是第 (i) 个 distractor proposal

• 因为自相关操作是线性的，则将(varphi(p_{k}))提出来：

[q=mathop{argmax}limits_{p_{k}inmathcal{P}}(varphi(z)-frac{hat{alpha}sum_{i=1}^{n}alpha_{i}f(d_{i},p_{k})} {sum_{i=1}^{n}alpha_{i}})*varphi(p_{k})]

3.Long term Tracking

• 当追丢时用local to global stategy 恒定step迭代的增加搜索区域的大小

to be continued