READING NOTE: Inside-Outside Net: Detecting Objects in Context with Skip Pooling and Recurrent Neura

原创 2016年06月01日 19:30:51

TITLE: Inside-Outside Net: Detecting Objects in Context with Skip Pooling and Recurrent Neural Networks

AUTHER: Sean Bell, C. Lawrence Zitnick, Kavita Bala, Ross Girshick Yan

ASSOCIATION: Cornell University, Microsoft Research

FROM: arXiv:1512.04143

CONTRIBUTIONS

  1. ION architecture is introduce that leverages context and multi-scale skip pooling for object detection. Use the information both inside and outside the ROI to determine the detection result.

METHOD

The main steps of the method is shown in the following figure.

  1. The image is first fed into a CNN, e.g.VGG16.
  2. ROI proposals are generated in the same way of Fast R-CNN.
  3. The information within the ROI are extracted by ROI pooling on different feature maps from different convolutional layers of different scales.
  4. The information outside the ROI are extracted by 2 successive 4-direction IRNNs. And ROI pooling is used to extract features.
  5. The pooled features are L2 nomalized and concated. Then a 1X1 conv layer is used to reduce the dimension.
  6. Two branches are learned to predict category and location.

some details

A 4-direction IRNN contains 4 independent IRNNs and each IRNN moves in different directions (left, right, up and down). The internal IRNN computations are splitted into separate logical layers. the input-to-hidden transition is implemented by a 1x1 convolution, and its computation can be shared across different directions.

ADVANTAGES

  1. The proposed detector works better on smaller objects compared with other works.
  2. Both local and global information are take into account.
  3. Skip pooling uses the informaiton of different scales.
  4. Two successive 4-direction IRNN cover the information form the whole image.
版权声明:本文为博主原创文章,未经博主允许不得转载。欢迎访问博主个人主页 https://joshua19881228.github.io/

论文《Inside-Outside Net: Detecting Objects in Context with skip pooling and Recurrent Neural Networks》

收录于CVPR20161.Contribution本文主要关注目标检测过程中的上下文信息以及多尺度信息两个方面,提出了Inside-Outside Net: Inside Net: 采用skipin...
  • yaoqi_isee
  • yaoqi_isee
  • 2017年03月19日 14:11
  • 2062

多尺度R-CNN(2): Inside-Outside Net: Detecting Objects in Context with Skip Pooling and Recurrent Neural

CNN高层特征具有丰富的语义信息,低层特征具有较高空间分辨率,研究如何融合不同层之间的特征,是物体检测领域热门的方向。近期很多工作通过融合多层来提升检测和分割的性能,按照融合与预测的先后顺序,分类为早...
  • Solomon1558
  • Solomon1558
  • 2017年05月31日 21:50
  • 950

目标检测2015

本文转载自:http://blog.csdn.net/zhuiqiuk/article/details/53613879 https://handong1587.github.io/deep_l...
  • csuzhaoqinghui
  • csuzhaoqinghui
  • 2017年03月02日 20:52
  • 926

综述:计算机视觉中RNN应用于目标识别

深度学习在计算机视觉领域取得的巨大的发展,最近几年CNN一直是目前主流模型所采取的架构。最近半年RNN/LSTM应用在识别领域逐渐成为一种潮流,RNN在获取目标的上下文中较CNN有独特的优势。以下我们...
  • yaoxingfu72
  • yaoxingfu72
  • 2016年07月03日 09:54
  • 5731

目标检测领域 2015

本文转载自:http://blog.csdn.net/zhuiqiuk/article/details/53613879 https://handong1587.github.io/deep_l...
  • hx921123
  • hx921123
  • 2017年02月19日 12:51
  • 5856

All of Recurrent Neural Networks (RNN)

— notes for the Deep Learning book, Chapter 10 Sequence Modeling: Recurrent and Recursive Nets. M...
  • omnispace
  • omnispace
  • 2017年11月01日 16:08
  • 218

Motion-Based Multiple Object Tracking

%% Motion-Based Multiple Object Tracking % This example shows how to perform automatic detection and...
  • lcj1105
  • lcj1105
  • 2015年11月27日 22:20
  • 1760

运动相机检测无人机-- Detecting Flying Objects using a Single Moving Camera

Detecting Flying Objects using a Single Moving Camera PAMI 2017 http://cvlab.epfl.ch/research/unma...
  • zhangjunhit
  • zhangjunhit
  • 2017年09月21日 09:40
  • 255

kettle_删除“共享输出表”引发的错误

原创作品,出自 “深蓝的blog” 博客,欢迎转载,转载时请务必注明出处,否则追究版权法律责任。 深蓝的blog:   kettle报错:Unexpected problem reading shar...
  • huangyanlong
  • huangyanlong
  • 2015年04月01日 08:48
  • 2032

Learnable pooling with Context Gating for video classification

这篇论文是2016年Google Cloud & YouTube-8M Video Understanding Challenge比赛中冠军得主的论文。 文章的两点贡献: 融合了VLAD, bag-...
  • Cheese_pop
  • Cheese_pop
  • 2017年10月09日 21:29
  • 365
内容举报
返回顶部
收藏助手
不良信息举报
您举报文章:READING NOTE: Inside-Outside Net: Detecting Objects in Context with Skip Pooling and Recurrent Neura
举报原因:
原因补充:

(最多只允许输入30个字)