作者：Bin Yan, Yi Jiang, Peize Sun, Dong Wang, Zehuan Yuan, Ping Luo, Huchuan Lu

**作者单位：**大连理工大学信息与通信工程学院，字节跳动，香港大学，彭成实验室

发布时间：2022

发布期刊/会议：Arxiv

出版商：Arxiv

论文全称：Towards Grand Unification of Object Tracking

论文地址：

Towards Grand Unification of Object Tracking

论文代码：

https://github.com/MasterBin-IIAU/Unicorn

地位：

个人理解

• 创新点：
  • 提出了一个统一的方法，使用一个统一的网络架构来解决目标跟踪中四类子问题（SOT，VOS，MOT，MOTS）
• 为什么：
  • 由于目标跟踪问题本身定义的碎片化，现有的大多数跟踪器都是针对单个或部分任务而开发的，并且对特定任务的特征过于专门化。
• 怎么做：
  1. 使用了共享权重的骨干网络提取当前帧和相关帧的视觉特征
  2. 使用**变形注意力模块对视觉特征提取Embedding特征，**接着计算像素间和实例间的相关性
  3. 设计 Target Prior，来满足 SOT、VOS、MOT、MOTS这4个任务的统一

一、摘要

We present a unified method, termed Unicorn, that can simultaneously solve four tracking problems (SOT, MOT, VOS, MOTS) with a single network using the same model parameters. Due to the fragmented definitions of the object tracking problem itself, most existing trackers are developed to address a single or part of tasks and overspecialize on the characteristics of specific tasks. By contrast, Unicorn provides a unified solution, adopting the same input, backbone, embedding, and head across all tracking tasks. For the first time, we accomplish the great unification of the tracking network architecture and learning paradigm. Unicorn performs on-par or better than its task-specific counterparts in 8 tracking datasets, including LaSOT, TrackingNet, MOT17, BDD100K, DA VIS16-17, MOTS20, and BDD100K MOTS. We believe that Unicorn will serve as a solid step towards the general vision model.