2019.1.7

一、Contour Detection and Hierarchical Image Segmentation

简单步骤概括:

算法路线:gPb-------OWT------UCM

每一部份的功能:

1. gPb(global Pb)计算每一个pixel的作为boundary的可能性,即pixel的weight;

2. OWT(Oriented Watershed Transform)将上述gPb的结果转换为多个闭合的regions;

3. UCM(Ultrametric Contour Map)将上述regions集,转换为hierarchical tree;
--------------------- 
作者:罗达志 
来源:CSDN 
原文:https://blog.csdn.net/Alex_luodazhi/article/details/47337327 

作用:为了产生用于 object detection的proposals

二、MCG(Multiscale Combinatorial Grouping)

代码:https://github.com/jponttuset/mcg

https://blog.csdn.net/skye1221/article/details/75125897

理解:https://www.cnblogs.com/hrhguanli/p/4841846.html

https://www.cnblogs.com/ariel-dreamland/p/8404696.html

MCG概述

三、 CPMC: Automatic object segmentation using constrained parametric min-cuts. 

J. Carreira and C. Sminchisescu. CPMC: Automatic object segmentation using constrained parametric min-cuts. PAMI,
34(7):1312–1328, 2012. 1, 2, 6, 7

来自:http://www.docin.com/p-1692947344.html 

四、Object Instance Segmentation and Fine-Grained Localization Using Hypercolumns

https://ieeexplore.ieee.org/document/7486965

五、

©️2020 CSDN 皮肤主题: 书香水墨 设计师:CSDN官方博客 返回首页