pytorch+net链接整理 2020-08-10

FasterR-CNN，R-FCN，SSD，FPN，RetinaNet,YOLOv3速度和准确性比较

https://www.jianshu.com/p/0a6875fec23a

卷积神经网络之LeNet（详细）

https://www.cnblogs.com/wangguchangqing/p/10329402.html

PyTorch中文文档

https://pytorch-cn.readthedocs.io/zh/latest/

视频：
https://www.bilibili.com/video/BV12741177Cu

深度学习目标检测综述

https://zhuanlan.zhihu.com/p/98756890?from_voters_page=true

吴恩达网络与深度学习第三周作业 - 带有一个隐藏层的平面数据分类

https://blog.csdn.net/u013733326/article/details/79702148

全部课程传送门

mobilenet介绍

https://www.jianshu.com/p/854cb5857070

darknet介绍

https://blog.csdn.net/u010122972/article/details/83541978

pytorch实际操作练习1

https://www.cnblogs.com/wj-1314/p/9830950.html

目标检测算法之YOLO系列算法的Anchor聚类

https://blog.csdn.net/just_sort/article/details/103386047

Windows10下使用OpenVINO需要手动编译cpu_extension.lib

https://www.jianshu.com/p/32d12abc6e6a

边框回归(Bounding Box Regression)详解（较为详细）

https://blog.csdn.net/zijin0802034/article/details/77685438

MTCNN工作原理

https://blog.csdn.net/qq_36782182/article/details/83624357
网络地址：
https://github.com/imistyrain/MTCNN/blob/master/Fast-MTCNN/model/det1.prototxt
https://github.com/imistyrain/MTCNN/blob/master/Fast-MTCNN/model/det2.prototxt
https://github.com/imistyrain/MTCNN/blob/master/Fast-MTCNN/model/det3.prototxt

大话CNN经典模型：GoogLeNet（从Inception v1到v4的演进）

https://my.oschina.net/u/876354/blog/1637819

UNET论文解读及网络结构

论文解读：https://www.jianshu.com/p/f9b0c2c74488

网络结构：https://github.com/warden3344/unet

VGG论文解读及网络结构

论文解读：https://blog.csdn.net/qq_40027052/article/details/79015827
网络结构：https://github.com/davidgengenbach/vgg-caffe/blob/master/model/VGG_ILSVRC_19_layers_deploy.prototxt

segnet网络结构

https://github.com/alexgkendall/SegNet-Tutorial/tree/master/Models

MobileNet_ssd原理

https://blog.csdn.net/jiang_ming_/article/details/82356642

12篇文章带你逛遍主流分割网络

https://zhuanlan.zhihu.com/p/68456636

https://zhuanlan.zhihu.com/p/63337839

【图像分割模型】从FCN说起

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649032209&idx=2&sn=ee63ef7fc1ac163cea8277b9a1dcaf84&chksm=8712b86cb065317aa4a79c774614965318a95333952ef31d9b6aaddfae0666d714bc0026cb8f&scene=21#wechat_redirect

fcn开源代码

github下载地址https://github.com/shelhamer/fcn.berkeleyvision.org

【图像分割模型】编解码结构SegNet

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649032348&idx=1&sn=80825e10fcfd502e2145ed9ce8e6b430&chksm=8712b8e1b06531f7723c37851204ccea66b212b24cf36315a2859aec07dbe0938c9a6bfe59fb&scene=21#wechat_redirect

segnet开源代码

https://github.com/alexgkendall/caffe-segnet

【分割模型解读】感受野与分辨率的控制术—空洞卷积

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649032510&idx=1&sn=e65528e6ce1d0c31d9c7f20cdb171a90&chksm=8712b943b0653055fe820db0fb56b87d7ef4032e82261914e437e3c5bcec59d6abfd930f7e1d&scene=21#wechat_redirect

【图像分割模型】快速道路场景分割—ENet

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649032625&idx=1&sn=48c542f5b6eab6b4f2fcaa515f503dc5&chksm=8712b9ccb06530da4afd439ef839d08e14cb835d5734e1c10ea4e73c03c531b2715c47ffd180&scene=21#wechat_redirect

【图像分割模型】以RNN形式做CRF后处理—CRFasRNN(学习的不是太透彻)

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649032798&idx=2&sn=12c1a4bad1fc6e323f527610206ab399&chksm=8712b623b0653f35d7f415fc2acdf7b8ae1a55b662b6b946881a52dd1e85ab3e9da0e781046a&scene=21#wechat_redirect

【图像分割模型】多感受野的金字塔结构—PSPNet

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649033063&idx=2&sn=0968ddb92b734671f6d96370cff872e9&chksm=8712b71ab0653e0c9edca8c119e58ec8082e19c2b19e9e8b03938c543e697244d480c39f6356&scene=21#wechat_redirect

【图像分割模型】全局特征与局部特征的交响曲—ParseNet

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649033178&idx=1&sn=d10da9a4784bbeffaa1de1805f5d3196&chksm=8712b7a7b0653eb112535f20bfdc967bdd948bf027feea563d1adfae44d120c9a815130c855a&scene=21#wechat_redirect

【图像分割模型】多分辨率特征融合—RefineNet

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649033325&idx=1&sn=b614ca880971265b3ea4f0ddf219fef6&chksm=8712b410b0653d0622982b05043beb4fe46372a3360b96d1354cebeb6546fb77a0f2f7b0ae01&scene=21#wechat_redirect

【图像分割模型】用BRNN做分割—ReSeg

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649033342&idx=1&sn=046270ce9f9816005453fb595e0d77c3&chksm=8712b403b0653d153ba01d1b0de8c770b1d6e22e13d4e21c9a8f3a5e440990c8721da5195720&scene=21#wechat_redirect

【图像分割模型】BRNN下的RGB-D分割—LSTM-CF

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649033419&idx=1&sn=e58110205016b96f30fda49fe82c177a&chksm=8712b4b6b0653da0f7b9955de358ec24bb2c707997ec3560f45742ea9448e22826111a4c0fc9&scene=21#wechat_redirect

【图像分割模型】实例分割模型—DeepMask

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649033459&idx=1&sn=db864ce462b594dc0e87bcc938dcb061&chksm=8712b48eb0653d98242f42c6df5f220518088f6a794fc7d63ac2489cb27a8178c1441588daa1&scene=21#wechat_redirect

图像分割综述

https://cloud.tencent.com/developer/article/1460945

分割网络（实时检测）

https://github.com/PINTO0309/MobileNet-SSD-RealSense/blob/master/caffemodel/SemanticSegmentation/512x1024_citysapes/deploy.prototxt

如何轻松愉快地理解条件随机场（CRF）？

https://blog.csdn.net/dcx_abc/article/details/78319246

ENET论文解析

https://blog.csdn.net/dcx_abc/article/details/78319246

ENET网络结构

深度学习论文笔记（六）— FCN-2015年（Fully Convolutional Networks for Semantic Segmentation）

https://blog.csdn.net/teeyohuang/article/details/75933339

【深度学习】深入理解Batch Normalization批标准化

https://www.cnblogs.com/guoyaohua/p/8724433.html

【论文翻译】SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation

https://blog.csdn.net/u014451076/article/details/70741629

Fully Convolutional Networks for semantic Segmentation（深度学习经典论文翻译）

https://www.cnblogs.com/xuanxufeng/p/6249834.html

面向新手的CNN入门指南（一）

https://zhuanlan.zhihu.com/p/37146355

ResNet-论文解读

https://blog.csdn.net/iModel/article/details/80688394

ResNet结构分析

https://blog.csdn.net/a940902940902/article/details/83858694

卷积的三种模式:full, same, valid

https://blog.csdn.net/leviopku/article/details/80327478

卷积神经网络(CNN)张量(图像)的尺寸和参数计算(深度学习)

https://www.cnblogs.com/touch-skyer/p/9150039.html

【图像分割模型】从FCN说起

https://mp.weixin.qq.com/s?__biz=MzA3NDIyMjM1NA==&mid=2649032209&idx=2&sn=ee63ef7fc1ac163cea8277b9a1dcaf84&chksm=8712b86cb065317aa4a79c774614965318a95333952ef31d9b6aaddfae0666d714bc0026cb8f&scene=21#wechat_redirect

【目标检测基础】Faster Rcnn框架原理解析

https://blog.csdn.net/wangpengfei163/article/details/80961275

faster-rcnn 之 RPN网络的结构解析

https://blog.csdn.net/sloanqin/article/details/51545125

（RegionProposal Network)RPN网络结构及详解

https://blog.csdn.net/qq_36269513/article/details/80421990

非极大值抑制(Non-Maximum Suppression)

非极大值抑制，简称为NMS算法，英文为Non-Maximum Suppression。其思想是搜素局部最大值，抑制极大值。NMS算法在不同应用中的具体实现不太一样，但思想是一样的。非极大值抑制，在计算机视觉任务中得到了广泛的应用，例如边缘检测、人脸检测、目标检测（DPM，YOLO，SSD，Faster R-CNN）等。

链接：https://www.jianshu.com/p/d452b5615850

深度学习常用的Data Set数据集和CNN Model总结

https://blog.csdn.net/qq_17448289/article/details/52850223

CNN中卷积层参数量与输出特征图(feature map)尺寸的计算公式

卷积层输入特征图(input feature map)的尺寸为：Hinput×Winput×Cinput
Hinput表示输入特征图的高
Winput表示输入特征图的宽
Cinput表示输入特征图的通道数（如果是第一个卷积层则是输入图像的通道数，如果是中间的卷积层则是上一层的输出通道数）
卷积层的参数有如下几个：
输出通道数为K
正方形卷积核的边长为F
步幅(stride)为S
补零的行数和列数(padding)为P
输出特征图(output feature map)的尺寸为Houtput×Woutput×Coutput，其中每一个变量的计算方式如下：
Houtput=(Hintput−F+2P)/S+1
Woutput=(Winput−F+2P)/S+1
Coutput=K
参数量大小的计算，分为weights和biases：
首先来计算weights的参数量：F×F×Cinput×K
接着计算biases的参数量：K
所以总参数量为：F×F×Cinput×K+K

深入理解AlexNet网络

https://blog.csdn.net/luoluonuoyasuolong/article/details/81750190

对抗神经网络学习和实现(GAN)

https://blog.csdn.net/just_sort/article/details/79454054

GAN学习指南：从原理入门到制作生成Demo

https://zhuanlan.zhihu.com/p/24767059

批量归一化

https://blog.csdn.net/leneey/article/details/80075356

【深度学习系列】卷积神经网络CNN原理详解(一)——基本原理

https://www.cnblogs.com/charlotte77/p/7759802.html

查看prototxt，在线工具

http://ethereon.github.io/netscope/#/editor

手把手教你用PyTorch从零搭建图像分类模型

https://blog.csdn.net/gaotihong/article/details/80763813

Python之Numpy库常用函数大全（含注释）

https://www.cnblogs.com/TensorSense/p/6795995.html

Yolov3+windows10+VS2015部署安装

(https://blog.csdn.net/sinat_26940929/article/details/80342660)

数据增强方法

反向传播算法的直观理解

https://blog.csdn.net/mao_xiao_feng/article/details/53048213

卷积神经网络CNN理解

https://blog.csdn.net/stdcoutzyx/article/details/41596663

【目标检测】Faster RCNN算法详解

http://blog.csdn.net/shenxiaolu1984/article/details/51152614#fn:1