码迷,mamicode.com
首页 > 其他好文 > 详细

[paper reading] C-MIL: Continuation Multiple Instance Learning for Weakly Supervised Object Detection CVPR2019

时间:2019-04-15 00:36:05      阅读:332      评论:0      收藏:0      [点我收藏+]

标签:generate   app   generated   multi   ati   overlap   无法   子集   paper   

MIL陷入局部最优,检测到局部,无法完整的检测到物体。将instance划分为空间相关和类别相关的子集。在这些子集中定义一系列平滑的损失近似代替原损失函数,优化这些平滑损失。

C-MIL learns instance subsets, where the instances are spatially related, i.e., overlapping with each other, and class related, i.e., having similar object class scores.

技术图片

C-MIL treats images as bags and image regions generated by an object proposal method [24,32] as instances

技术图片

 

[paper reading] C-MIL: Continuation Multiple Instance Learning for Weakly Supervised Object Detection CVPR2019

标签:generate   app   generated   multi   ati   overlap   无法   子集   paper   

原文地址:https://www.cnblogs.com/demian/p/10668677.html

(0)
(0)
   
举报
评论 一句话评论(0
登录后才能评论!
© 2014 mamicode.com 版权所有  联系我们:gaon5@hotmail.com
迷上了代码!