[CVPR2017] Weakly Supervised Cascaded Convolutional Networks论文笔记

news/2024/10/9 11:38:57/

https://www.csee.umbc.edu/~hpirsiav/papers/cascade_cvpr17.pdf

Weakly Supervised Cascaded Convolutional Networks, Ali Diba, Vivek Sharma, Ali Pazandeh, Hamed Pirsiavash and Luc Van Gool

  

亮点

  • 通过多任务叠加(分类,分割)提高了多物体弱监督检测的正确率
  • 通过利用segmentation筛选纯净的proposals,得到了更鲁棒的结果
  • 为弱监督分割任务设计比较鲁棒的loss
    • 只考虑全局的分类结果和置信度对高的部分
    • 通过loss的weights关注到最需要关注的部分

相关工作 

 

One of the most common approaches [7] consists of the following steps:

 

  • generates object proposals,
  • extracts features from the proposals,
  • applies multiple instance learning (MIL) to the features and finds the box labels from the weak bag (image) labels. 

弱监督物体检测难点: 弱监督物体检测对初始化要求很高,不好的初始化可能会使网络陷入局部最优解,解决的办法主要有以下几个:

  • improve the initialization [31, 9, 28, 29]
  • regularizing the optimization strategies [4, 5, 7]
  • [17] employ an iterative self-learning strategy to employ harder samples to a small set of initial samples
  • [15] use a convex relaxation of soft-max loss 

Majority of the previous works [25, 32] use a large collection of noisy object proposals to train their object detector. In contrast, our method only focuses on a very few clean collection of object proposals that are far more reliable, robust, computationally efficient, and gives better performance

方法

Two-stage: proposal and image classification (conv1 till con5, global pooling) + multiple instance learning (2fc, score layer)

 

 

1. image classification: CNN with global average pooling (GAP) [36]中引入,将分类过程中fc层的weights作为原来convolutional layer输出的权重并将所有频道加权得到的图作为class activation map。在这一步中,还产生一个分类的loss LGAP

[36]  B. Zhou, A. Khosla, A. Lapedriza, A. Oliva, and A. Torralba. Learning deep features for discriminative localization. In CVPR, 2016. 3, 4, 5, 6, 7, 8

 

 

2. multiple instance learning

Proposal: edgeboxs [37] is used to generate an initial set of object proposals. Then we threshold the class activation map [36] to come up with a mask. Finally, we choose the initial boxes with largest overlap with the mask.

 

 

Three-stage:  more information about the objects’ boundary learned in a segmentation task can lead to acquisition of a better appearance model and then better object localization.

  • 主要思想:分割监督信号帮助提升定位准确率。
  • 弱分割监督信号:上一级得到的mask

 

实验结果

 

PASCAL VOC 2007

  • +3.3% classification compared with [18]
  • +1.6% correct localization compared with [27]
  • +0.6% compared with [6]

PASCAL VOC 2010

  • +3.3% compared with [6]

PASCAL VOC 2012

  • +8.8% compared with [18]
  • ILSVRC 2013
  • +5.5% compared with [18]

Object detection training

  • PASCAL VOC 2007 test set: Faster RCNN trained by the pseudo ground-truth (GT) bounding boxes generated by our cascaded networks performs slightly better than our transfered model. (+0.3%)

[6] H. Bilen and A. Vedaldi. Weakly supervised deep detection networks. In CVPR, 2016. 6, 7, 8

[18] D. Li, J.-B. Huang, Y. Li, S. Wang, and M.-H. Yang. Weakly supervised object localization with progressive domain adaptation. In IEEE Conference on Computer Vision and Pattern Recognition, 2016. 2, 6, 7

[27] K. Simonyan and A. Zisserman. Very deep convolutional networks for large-scale image recognition. In ICLR, 2015. 5, 6

转载于:https://www.cnblogs.com/Xiaoyan-Li/p/8651579.html


http://www.ppmy.cn/news/323171.html

相关文章

蛇行矩阵-C语言

题目 蛇形矩阵是由1开始的自然数依次排列成的一个矩阵上三角形。 输入格式: 本题有多组数据,每组数据由一个正整数N组成。(N不大于100) 输出格式: 对于每一组数据,输出一个N行的蛇形矩阵。 两组输出之…

模拟数字电路->绪论

模拟数字电路 绪论信号相关放大电路模型抽象放大器 放大电路模型分析电压放大电路电流放大模型 主要性能指标增益Av,输入电阻Ri,输出电阻Ro频率响应和非线性失真非线性失真 问题总结题型输出电压,输出电阻放大倍数 课程无关 绪论 信号相关 信号:信息的载体 电信号源的电路表达…

leetcode链表(简单难度)

leetcode链表简单题 21.合并两个有序链表83.删除排序链表中的重复元素141.环形链表160.相交链表暴力法长度差法通过“增加”消除长度差 203.移除链表元素206.反转链表234.回文链表237.删除链表中的节点876.链表的中间节点1290.二进制链表转整数get flag 21.合并两个有序链表 /…

机械制造作业考研题目答案分享——金属切削规律2

金属减削规律的四道题目以及答案解释(english version)。 版权声明 本内容由狂小虎原创整合,请不要售卖,为了防止爬虫以及保持免费性,设置为仅粉丝可见。另外,题目以及解释可能不完全正确,仅供参考,同时也…

小白C语言Leetcode————160.相交链表

本题我按Leetcode上所要求的函数运行出错,但我大概想明白在何处出错:创造两个链表时时分别创造的,所以在创造这两个链表时他们占用了不同的地址空间,即使他们的值相同,但他们并不相交,因此出错。 因此在这…

java翻译smali_求大佬帮我把这个smali文件翻译成java或者C#好吗

[Asm] 纯文本查看 复制代码.class public Lcom/mob/commons/a; .super Ljava/lang/Object; # static fields .field private static a:Ljava/util/HashMap; .annotation system Ldalvik/annotation/Signature; value { "Ljava/util/HashMap", " "Ljava/la…

AE+VS+c#开发颜色符号系统之点值符号化(七)

首先总结一下之前写的几篇符号化文章,列表如下: AEVS开发颜色符号系统之单一值符号(一) AEVS开发颜色符号系统之唯一值符号(二) AEVS开发颜色符号系统之分类符号(三) AEVS开发颜…

Geant 4 生成 ROOT 文件(初学)

exmapleB4b 源代码 B4bRunAction.cc(部分) 1. 在 RunAction() 构造函数中,定义 直方图或TTree 的格式 // Create analysis manager// The choice of analysis technology is done via selectin of a namespace// in B4Analysis.hhauto analysisManager G4Analysis…