How much Position Information Do Convolutional Neural Networks Encode?

知乎上看到一篇博客https://zhuanlan.zhihu.com/p/99766566，讨论cnn中位置信息是如何学到的，感觉挺有意思的，就看了一下原文，

简单记录一下几个实验部分。

1、位置信息来自预训练模型

　作者分别用预训练权重的vgg,resnet和posenet来对比，发现用了预训练权重的vgg和resnet提取的位置信息更准确。但这个实验我

觉得设计的不合理，真正对比实验应该设计的是，用预训练权重的vgg，resnet对比随机初始化的vgg,resnet，而不是和单层的posnet来对比

2、大的卷积核和更多的层数，能获得更准确的位置信息

更多的层数就不用说了，更大的卷积核也符合light head rcnn中得到的结论的，在某层中使用大的卷积核，效果更好。

3、位置信息存放在那一层

原文分别用5层特征来对比，最后发现越深的特征，结果越准确。

4、位置信息来自哪里

　这篇文章的核心，作者认为来自于padding，并设计了几组对比实验（包括检测分割任务）：加padding的准确率比不加的准确率要高很多。

原文地址：https://www.cnblogs.com/573177885qq/p/12243233.html

时间： 2024-11-02 11:48:47

How much Position Information Do Convolutional Neural Networks Encode?的相关文章

[论文理解] How much position information do convolutional neural network encode?

How much position information do convolutional neural network encode? Intro 文章是ICML2020的一个工作,探究了CNN到底有没有编码位置信息,这些位置信息在哪些神经元中被编码.这些位置信息又是如何被暴露给神经网络学习的.文章通过大量实验表明,CNN不仅可以编码位置信息,而且越深的层所包含的位置信息越多(而往往越深的层解释性越差,浅层学习到的形状.边缘等比较容易解释),而位置信息是通过zero-padding透露的,显

（转）A Beginner's Guide To Understanding Convolutional Neural Networks

Adit Deshpande CS Undergrad at UCLA ('19) Blog About A Beginner's Guide To Understanding Convolutional Neural Networks Introduction Convolutional neural networks. Sounds like a weird combination of biology and math with a little CS sprinkled in, but

卷积神经网络用于视觉识别Convolutional Neural Networks for Visual Recognition

Table of Contents: Architecture Overview ConvNet Layers Convolutional Layer Pooling Layer Normalization Layer Fully-Connected Layer Converting Fully-Connected Layers to Convolutional Layers ConvNet Architectures Layer Patterns Layer Sizing Patterns C

A Beginner's Guide To Understanding Convolutional Neural Networks(转)

A Beginner's Guide To Understanding Convolutional Neural Networks Introduction Convolutional neural networks. Sounds like a weird combination of biology and math with a little CS sprinkled in, but these networks have been some of the most influential

Image Scaling using Deep Convolutional Neural Networks

Image Scaling using Deep Convolutional Neural Networks This past summer I interned at Flipboard in Palo Alto, California. I worked on machine learning based problems, one of which was Image Upscaling. This post will show some preliminary results, dis

Convolutional Neural Networks from deep learning (assignment 1 from week 1)

Convolutional Neural Networks https://www.coursera.org/learn/convolutional-neural-networks/home/welcome There are still something confuse me! working how to paste the jupyter here.................................................... Convolutional Neur

[C6] Andrew Ng - Convolutional Neural Networks

About this Course This course will teach you how to build convolutional neural networks and apply it to image data. Thanks to deep learning, computer vision is working far better than just two years ago, and this is enabling numerous exciting applica

Learning to Compare Image Patches via Convolutional Neural Networks --- Reading Summary

Learning to Compare Image Patches via Convolutional Neural Networks --- Reading Summary 2017.03.08 Target: this paper attempt to learn a geneal similarity function for comparing image patches from image data directly. There are several ways in which

（转）A Beginner's Guide To Understanding Convolutional Neural Networks Part 2

Adit Deshpande CS Undergrad at UCLA ('19) Blog About A Beginner's Guide To Understanding Convolutional Neural Networks Part 2 Introduction Link to Part 1 In this post, we’ll go into a lot more of the specifics of ConvNets. Disclaimer: Now, I do reali