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 http://joshua881228.webfactional.com/ 举报

相关文章推荐

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...

用 Python 和 OpenCV 检测图片上的条形码Detecting Barcodes in Images with Python and OpenCV

UPDATE: The introduction to this post may seen a little “out there”. For some context, I had just fi...

我是如何成为一名python大咖的?

人生苦短,都说必须python,那么我分享下我是如何从小白成为Python资深开发者的吧。2014年我大学刚毕业..

【ECCV2016】Detecting Text in Natural Image with Connectionist Text Proposal Network

本文的方法最大亮点在于把RNN引入检测问题(以前一般做识别)。文本检测,先用CNN得到深度特征,然后用固定宽度的anchor来检测text proposal(文本线的一部分),并把同一行anchor对...

Context-aware Natural Language Generation with Recurrent Neural Networks

《Context-aware Natural Language Generation with Recurrent Neural Networks》 使用循环神经网络的上下文感知自然语言生成 1.背景...

[论文复现]Detecting Text in Natural Image with Connectionist Text Proposal Network

本文转载自: http://blog.csdn.net/peaceinmind/article/details/53215090 最近一直在复现这篇论文,除了数据外已基本完成,...

[论文复现]Detecting Text in Natural Image with Connectionist Text Proposal Network

最近一直在复现这篇论文,除了数据外已基本完成,可是没想到昨天开源了[code],只能说我等水货赶不上开源的速度,但是又不开源数据,然后默默地发现,做深度学习这等东西,代码不重要,重要的是数据,以后说不...

Inside The C++ Object Model(Note):sizeof in an object

Data语义学:一.一个类得实际大小受到三个因素影响:1.语言本身所造成的额外负担.比如 virtual ,虚基类.2.编译器对于特殊情况所提供的优化处理3.内存对齐(Alignment)的限制
返回顶部
收藏助手
不良信息举报
您举报文章:深度学习:神经网络中的前向传播和反向传播算法推导
举报原因:
原因补充:

(最多只允许输入30个字)