基于深度学习的目标跟踪

先看看总结性的,写的都挺好

多目标跟踪全解析,全网最全 (非常好)

工程实践中,目标检测为何要加目标追踪?

再谈多目标追踪--快速应用部署

MOT榜前算法探讨

Anchor-free应用一览:目标检测、实例分割、多目标跟踪

Tracking Objects as Points

Github:GitHub - xingyizhou/CenterTrack: Simultaneous object detection and tracking using center points.

基于CenterNet如何评价最新的两篇基于point的目标检测模型CenterNet?

CenterNet:Objects as Points论文学习笔记+代码复现

写的特别好:论文精读——CenterNet :Objects as Points

代码解释特别好:centernet理解+demo代码分析

CenterTrack论文学习笔记 - CenterTrack

有关CenterTrack的一些注意点

基于CenterTrack的3D目标检测源码解读

缺点也是优点:

    CenterTrack is purely local. It only associates objects in adjacent frames, with-out reinitializing lost long-range tracks. It trades the ability to reconnect long-range tracks for simplicity, speed, and high accuracy in the local regime.

A Simple Baseline for Multi-Object Tracking

Paper:https://arxiv.org/abs/2004.01888

Github:GitHub - ifzhang/FairMOT: [IJCV-2021] FairMOT: On the Fairness of Detection and Re-Identification in Multi-Object Tracking

打遍天下无敌手,却说它只是个baseline!多目标跟踪FairMOT的烦恼 - 云+社区 - 腾讯云

论文阅读-FairMOT:《A Simple Baseline for Multi-Object Tracking》

FairMOT理解与实现

FairMOT项目代码实战,谦称青铜,实为王者

CVPR 2020 多目标跟踪算法 FairMOT代码demo运行及训练

Detection计算人的类间特征,区分检测部分是人还是背景。Re-ID计算人的类内特征,用来区分这个人到底是哪个人。

ByteTrack

Github:

https://github.com/ifzhang/ByteTrack

作者讲解

比较

目标跟踪(二) SDE, JDE, FairMot, CenterTrack, Bytetrack 跟踪比较​​​​​​​

猜你喜欢

转载自blog.csdn.net/qq_36076110/article/details/106669011