当前位置:网站首页>Semantic segmentation | learning record (5) FCN network structure officially implemented by pytoch
Semantic segmentation | learning record (5) FCN network structure officially implemented by pytoch
2022-07-08 02:10:00 【coder_ sure】
List of articles
Preface
Pytorch Officially realized FCN It is slightly different from the structure diagram of the paper in that year , Because now there are more backbone The choice of , And the expansion convolution technology is applied .
FCN Network structure
The picture below is Pytorch Officially realized FCN Network structure :
The picture below is Resnet Network structure :
Compare the two networks for analysis :FCN Online backbone The choice is Resnet50 backbone
, In this part of the blue dotted box in front of the structure and Resnet It's exactly the same .
Focus on layer3 and layer4 Part of .FCN Of layer3 and layer4 Corresponding to Resnet Of conv4_x and conv5_x. The difference lies in :
- Resnet Of
conv4_x
Medium 6 The residual structures correspond to FCN Oflayer3
Medium 1 individualBottleneck1
and 5 individualBottleneck2.
- Resnet Of
conv5_x
Medium 3 The residual structures correspond to FCN Oflayer4
Medium 1 individualBottleneck1
and2 individual Bottleneck2
.
Let's focus on that
Bottleneck1
andBottleneck2
Structure :
Bottleneck1
Corresponding to the dotted line structure in the residual structure , Here and resnet The difference is that it will be a shortcut to branch on 1 ∗ 1 1*1 1∗1 ConvolutionThe step length is changed to 1.
Therefore, there is no down sampling operation here , If the down sampling ratio in semantic segmentation is too large, the original image effect will be affected . In addition to 3 ∗ 3 3*3 3∗3 At the convolution kernelThe step length is also changed to 1, And the expansion convolution is introduced
.Bottleneck2
It also introducesExpansion convolution .
Let's take a look at FCN Head part :
- Through a 3*3 The convolution of layer , Enter the... Of the characteristic graph channel The number is adjusted to the original 1/4 Turn into 512.
- Through one Dropout layer
- Through one more 1*1 Convolution layer , Adjust the feature layer channel For the number of categories .
- Finally, through a bilinear interpolation method , Return to the original size .
There is another one on the right FCN Head, The official reasons : Preventing error gradients cannot be transferred to the shallow layer of the network
, So the auxiliary classifier is introduced here .
You can enable this auxiliary classifier during training , You can try it . When actually deploying to the real environment , We only use output.
Reference material
边栏推荐
- 魚和蝦走的路
- JVM memory and garbage collection-3-runtime data area / method area
- SQLite3 data storage location created by Android
- Ml self realization / logistic regression / binary classification
- Reading notes of Clickhouse principle analysis and Application Practice (7)
- Nmap tool introduction and common commands
- burpsuite
- How to make the conductive slip ring signal better
- What are the types of system tests? Let me introduce them to you
- 银行需要搭建智能客服模块的中台能力,驱动全场景智能客服务升级
猜你喜欢
burpsuite
metasploit
VR/AR 的产业发展与技术实现
Application of slip ring in direct drive motor rotor
Talk about the realization of authority control and transaction record function of SAP system
Can you write the software test questions?
OpenGL/WebGL着色器开发入门指南
Usage of hydraulic rotary joint
Little knowledge about TXE and TC flag bits
微信小程序uniapp页面无法跳转:“navigateTo:fail can not navigateTo a tabbar page“
随机推荐
The function of carbon brush slip ring in generator
JVM memory and garbage collection-3-object instantiation and memory layout
Anan's judgment
PB9.0 insert OLE control error repair tool
给刚入门或者准备转行网络工程师的朋友一些建议
The method of using thread in PowerBuilder
微软 AD 超基础入门
Usage of hydraulic rotary joint
鱼和虾走的路
《ClickHouse原理解析与应用实践》读书笔记(7)
快手小程序担保支付php源码封装
Deeppath: a reinforcement learning method of knowledge graph reasoning
PHP calculates personal income tax
Towards an endless language learning framework
Nanny level tutorial: Azkaban executes jar package (with test samples and results)
Introduction à l'outil nmap et aux commandes communes
[target tracking] |dimp: learning discriminative model prediction for tracking
Ml self realization / linear regression / multivariable
WPF custom realistic wind radar chart control
谈谈 SAP 系统的权限管控和事务记录功能的实现