Paper Read: Robust Deep Multi-modal Learning Based on Gated Information Fusion Network
Robust Deep Multi-modal Learning Based on Gated Information Fusion Network
2018-07-27 14:25:26
Paper:https://arxiv.org/pdf/1807.06233.pdf
Related Papers:
1. Infrared and visible image fusion methods and applications: A survey Paper
2. Chenglong Li, Xiao Wang, Lei Zhang, Jin Tang, Hejun Wu, and Liang Lin. WELD: Weighted Low-rank Decomposition or Robust Grayscale-Thermal Foreground Detection. IEEE Transactions on Circuits and Systems for Video Technology (T-CSVT), 27(4): 725-738, 2017. [Project pagewith Dataset and Code]
3. Chenglong Li, Xinyan Liang, Yijuan Lu, Nan Zhao, and Jin Tang. RGB-T Object Tracking: Benchmark and Baseline.[arXiv] [Dataset: Google drive, Baidu cloud] [Project page]
本文针对多模态融合问题(Multi-modal),提出一种基于 gate 机制的融合策略,能够自适应的进行多模态信息的融合。作者将该方法用到了物体检测上,其大致流程图如下所示:

如上图所示,作者分别用两路 Network 来提取两个模态的特征。该网络是由标准的 VGG-16 和 8 extra convolutional layers 构成。另外,作者提出新的 GIF(Gated Information Fusion Network) 网络进行多个模态之间信息的融合,以取得更好的结果。动机当然就是多个模态的信息,是互补的,但是有的信息帮助会更大,有的可能就质量比较差,功效比较小,于是就可以自适应的来融合,达到更好的效果。
Gated Information Fusion Network (GIF):

如上图所示:
该 GIF 网络的输入是:已经提取的 CNN feature map,这里是 F1, F2. 然后,将这两个 feature 进行 concatenate,得到 $F_G$. 该网络包含两个部分:
1. information fusion network(图2,虚线框意外的部分);
2. weight generation network (WG Network,即:图2,虚线处);
Weight Generation Network 分别用两个 3*3*1 的卷积核对组合后的 feature map $F_G$ 进行操作,然后输入到 sigmoid 函数中,即:gate layer,然后输出对应的权重 $w_1$,$w_2$。
Information fusion network 分别用得到的两个权重,点乘原始的 feature map,得到加权以后的特征图,将两者进行 concatenate 后,用 1*1*2k 的卷积核,得到最终的 feature map。

总结整个过程,可以归纳为:

== Done !
Paper Read: Robust Deep Multi-modal Learning Based on Gated Information Fusion Network的更多相关文章
- Exploring Architectural Ingredients of Adversarially Robust Deep Neural Networks
目录 概 主要内容 深度 宽度 代码 Huang H., Wang Y., Erfani S., Gu Q., Bailey J. and Ma X. Exploring architectural ...
- 【论文简读】 Deep web data extraction based on visual
<Deep web data extraction based on visual information processing>作者 J Liu 上海海事大学 2017 AIHC会议登载 ...
- Paper List ABOUT Deep Learning
Deep Learning 方向的部分 Paper ,自用.一 RNN 1 Recurrent neural network based language model RNN用在语言模型上的开山之作 ...
- 【RS】Deep Learning based Recommender System: A Survey and New Perspectives - 基于深度学习的推荐系统:调查与新视角
[论文标题]Deep Learning based Recommender System: A Survey and New Perspectives ( ACM Computing Surveys ...
- [转]Deep Reinforcement Learning Based Trading Application at JP Morgan Chase
Deep Reinforcement Learning Based Trading Application at JP Morgan Chase https://medium.com/@ranko.m ...
- 论文笔记: Deep Learning based Recommender System: A Survey and New Perspectives
(聊两句,突然记起来以前一个学长说的看论文要能够把论文的亮点挖掘出来,合理的进行概括23333) 传统的推荐系统方法获取的user-item关系并不能获取其中非线性以及非平凡的信息,获取非线性以及非平 ...
- Predicting effects of noncoding variants with deep learning–based sequence model | 基于深度学习的序列模型预测非编码区变异的影响
Predicting effects of noncoding variants with deep learning–based sequence model PDF Interpreting no ...
- 论文翻译:2021_Towards model compression for deep learning based speech enhancement
论文地址:面向基于深度学习的语音增强模型压缩 论文代码:没开源,鼓励大家去向作者要呀,作者是中国人,在语音增强领域 深耕多年 引用格式:Tan K, Wang D L. Towards model c ...
- Deep High-Resolution Representation Learning for Human Pose Estimation
Deep High-Resolution Representation Learning for Human Pose Estimation 2019-08-30 22:05:59 Paper: CV ...
随机推荐
- hdu 5126 cdq+Treap+BIT
这题说的是给了三维空间然后操作 寻求在 x1,y1,z1 x2, y2, z2; (x1<x2, y1<y2,z1<z2) 计算出在 以这两个端点为右下和左上端点的方体内的点的 ...
- Solid Dominoes Tilings (轮廓线dp打表 + 容器)
第一步先打一个表,就是利用轮廓线DP去打一个没有管有没有分界线组合数量的表 #include<bits/stdc++.h> using namespace std; ; <<; ...
- python练习题-打印斐波拉契数列前n项
打印斐波拉契数列前n项 #encoding=utf-8 def fibs(num): result =[0,1] for i in range(num-2): result. ...
- GUI相应鼠标事件
function varargout = GUI18(varargin) % GUI18 MATLAB code for GUI18.fig % GUI18, by itself, creates a ...
- JXNU暑期选拔赛
最小的数 Time Limit : 3000/1000ms (Java/Other) Memory Limit : 65535/32768K (Java/Other) Total Submissi ...
- 前端框架VUE----cli脚手架(框架)
一.创建vue项目 npm install vue-cli -g #-g全局 (sudo)npm install vue-cli -g #mac笔记本 vue-init webpack myvue # ...
- 在centos上搭建JavaWeb环境(jdk+mysql+tomcat)
1.安装OpenJDK yum list java* -openjdk* -y java -version 2.安装Tomcat cd /usr/local wget https://mc.qclou ...
- bp暴力破解(转载)
在kali linux系统环境下自带burpsuite软件工具. 一.打开浏览器需要先设置将代理设置为本地. 打开firefox浏览器->open menu->preferences-&g ...
- php 网站中文简体繁体转换类
php 网站中文简体繁体转换类 <?php /* * define zh convert functions * 2017-4-28 use str_replace for speed * zh ...
- 【题解】Luogu P1204 [USACO1.2]挤牛奶Milking Cows
原题传送门:P1204 [USACO1.2]挤牛奶Milking Cows 实际是道很弱智的题目qaq 但窝还是觉得用珂朵莉树写会++rp(窝都初二了,还要考pj) 前置芝士:珂朵莉树 窝博客里对珂朵 ...