文件名称:Pedestrian-Detection
介绍说明－－下载内容均来自于网络，请自行研究使用
ICCV2013：
简 称UDN算法，从文中描述的检测效果来看，该方法是所有方法中最好的，并且，效果远超过其他方法。经过对论文和该算法源码的研究，该算法是与作者另外一篇 论文的方法 ，另外的论文算法做图片扫描，得到矩形框，然后用该方法对矩形框进行进一步确认，以及降低误警率和漏警率。另外的论文是：Multi-Stage Contextual Deep Learning for Pedestrian Detection
说得难听一点，这篇文章对行人检测没有多大的贡献。仅仅是用深度学习的CNN做candidate window的确认。而主要的行人检测的算法还是HOG+CSS+adaboost-ICCV2013:
UDN algorithm, described in the paper the detection results, the method is the best of all the methods, and the effect is far more than other methods. Through the research of the thesis and the source code of the algorithm, the algorithm is and author also a paper method, also the algorithm do scan pictures and get the rectangular box, then by the method of rectangular box for further confirmation, and reduce the false alarm rate and false alarm rate. Another paper is: Contextual Deep Learning for Pedestrian Multi-Stage Detection
To put it bluntly, this article does not have much contribution to the pedestrian detection. Just use deep learning s CNN to do window candidate s confirmation. And the main pedestrian detection algorithm is HOG+CSS+adaboost
简 称UDN算法，从文中描述的检测效果来看，该方法是所有方法中最好的，并且，效果远超过其他方法。经过对论文和该算法源码的研究，该算法是与作者另外一篇 论文的方法 ，另外的论文算法做图片扫描，得到矩形框，然后用该方法对矩形框进行进一步确认，以及降低误警率和漏警率。另外的论文是：Multi-Stage Contextual Deep Learning for Pedestrian Detection
说得难听一点，这篇文章对行人检测没有多大的贡献。仅仅是用深度学习的CNN做candidate window的确认。而主要的行人检测的算法还是HOG+CSS+adaboost-ICCV2013:
UDN algorithm, described in the paper the detection results, the method is the best of all the methods, and the effect is far more than other methods. Through the research of the thesis and the source code of the algorithm, the algorithm is and author also a paper method, also the algorithm do scan pictures and get the rectangular box, then by the method of rectangular box for further confirmation, and reduce the false alarm rate and false alarm rate. Another paper is: Contextual Deep Learning for Pedestrian Multi-Stage Detection
To put it bluntly, this article does not have much contribution to the pedestrian detection. Just use deep learning s CNN to do window candidate s confirmation. And the main pedestrian detection algorithm is HOG+CSS+adaboost
(系统自动生成,下载前可以参看下载内容)
下载文件列表
Joint Deep Learning for Pedestrian Detection\JDN_code\CNN\CDBNModel.mat
............................................\........\...\cnnapplygrads.m
............................................\........\...\cnnbp.m
............................................\........\...\cnnexamples.asv
............................................\........\...\cnnexamples.m
............................................\........\...\cnnff.m
............................................\........\...\CNNModel_init.mat
............................................\........\...\cnnsetup3.asv
............................................\........\...\cnnsetup3.m
............................................\........\...\cnntest.m
............................................\........\...\cnntrain.asv
............................................\........\...\cnntrain.m
............................................\........\...\compile.m
............................................\........\...\copycnnmodel.m
............................................\........\...\dtAccS.cc
............................................\........\...\dtAccS.mexw64
............................................\........\...\fconvn.cc
............................................\........\...\fconvn.mexw64
............................................\........\...\G.mat
............................................\........\...\GetAvgMiss.m
............................................\........\...\GetData_datareader.m
............................................\........\...\GetRegularizedW.m
............................................\........\...\GetSelWeight.m
............................................\........\...\showboxes.m
............................................\........\...\testCNNAll.m
............................................\........\...\testCNNCaltechTest2.m
............................................\........\...\testCNNCaltechTest4.asv
............................................\........\...\Testing.m
............................................\........\G.mat
............................................\........\model\CaltechTrain\CNN_CDBN_Model_iter2.mat
............................................\........\.....\INRIA\CNN_CDBN_Model_iter1.mat
............................................\........\.....\.....\CNN_CDBN_Model_iter2.mat
............................................\........\.....\.....\CNN_CDBN_Model_iter3.mat
............................................\........\.....\.....\CNN_CDBN_Model_iter4.mat
............................................\........\.....\.....\CNN_CDBN_Model_iter5.mat
............................................\........\NN\nnapplygrads.m
............................................\........\..\nnbp.m
............................................\........\..\nnchecknumgrad.m
............................................\........\..\nnexamples.m
............................................\........\..\nnff.m
............................................\........\..\nnsetup.m
............................................\........\..\nntest.m
............................................\........\..\nntrain.m
............................................\........\tmptoolbox\channels\chnsCompute.m
............................................\........\..........\........\chnsPyramid.m
............................................\........\..........\........\chnsScaling.m
............................................\........\..........\........\Contents.m
............................................\........\..........\........\convBox.m
............................................\........\..........\........\convMax.m
............................................\........\..........\........\convTri.m
............................................\........\..........\........\gradient2.m
............................................\........\..........\........\gradientHist.m
............................................\........\..........\........\gradientMag.m
............................................\........\..........\.