當前位置：首頁 > 编程语言 > asp.net >内容正文

asp.net

利用TinyXML读取VOC2012数据集的XML标注文件裁剪出所有人体目标保存为文件

發布時間：2025/3/21 asp.net 82 豆豆

生活随笔收集整理的這篇文章主要介紹了利用TinyXML读取VOC2012数据集的XML标注文件裁剪出所有人体目标保存为文件小編覺得挺不錯的,現在分享給大家,幫大家做個參考.

轉載自：利用TinyXML讀取VOC2012數據集的XML標注文件裁剪出所有人體目標保存為文件 - Why So Serious? - 博客頻道 - CSDN.NET ?http://blog.csdn.net/masibuaa/article/details/16104717

PASCAL VOC目標檢測數據集(The PASCAL Visual Object Classes)

http://pascallin.ecs.soton.ac.uk/challenges/VOC/

圖片中的目標用XML文件標注，格式為：

[html]?view plaincopy

<annotation>??

????<folder>VOC2012</folder>??

????<filename>2007_000346.jpg</filename>??

????<source>??

????????<database>The?VOC2007?Database</database>??

????????<annotation>PASCAL?VOC2007</annotation>??

????????<image>flickr</image>??

????</source>??

????<size>??

????????<width>500</width>??

????????<height>375</height>??

????????<depth>3</depth>??

????</size>??

????<segmented>1</segmented>??

????<object>??

????????<name>bottle</name>??

????????<pose>Unspecified</pose>??

????????<truncated>0</truncated>??

????????<difficult>0</difficult>??

????????<bndbox>??

????????????<xmin>124</xmin>??

????????????<ymin>107</ymin>??

????????????<xmax>230</xmax>??

????????????<ymax>343</ymax>??

????????</bndbox>??

????</object>??

????<object>??

????????<name>person</name>??

????????<pose>Unspecified</pose>??

????????<truncated>0</truncated>??

????????<difficult>0</difficult>??

????????<bndbox>??

????????????<xmin>137</xmin>??

????????????<ymin>78</ymin>??

????????????<xmax>497</xmax>??

????????????<ymax>375</ymax>??

????????</bndbox>??

????</object>??

????<object>??

????????<name>person</name>??

????????<pose>Unspecified</pose>??

????????<truncated>1</truncated>??

????????<difficult>0</difficult>??

????????<bndbox>??

????????????<xmin>89</xmin>??

????????????<ymin>202</ymin>??

????????????<xmax>129</xmax>??

????????????<ymax>247</ymax>??

????????</bndbox>??

????</object>??

????<object>??

????????<name>person</name>??

????????<pose>Frontal</pose>??

????????<truncated>1</truncated>??

????????<difficult>0</difficult>??

????????<bndbox>??

????????????<xmin>72</xmin>??

????????????<ymin>209</ymin>??

????????????<xmax>111</xmax>??

????????????<ymax>259</ymax>??

????????</bndbox>??

????</object>??

</annotation>??

對應的圖片為：

所以如果想用這個數據集做某種目標識別的訓練集的話，需要先從中裁出需要的目標。

下面這個程序就是這個目的，其中用到了TinyXML這個簡單易用的XML解析器(XML入門)

[cpp]?view plaincopy

#include?<iostream>??

#include?<fstream>??

#include?<opencv2/core/core.hpp>??

#include?<opencv2/highgui/highgui.hpp>??

#include?<opencv2/imgproc/imgproc.hpp>??

#include?<opencv2/objdetect/objdetect.hpp>??

#include?<opencv2/ml/ml.hpp>??

#include?<tinyxml.h>??

using?namespace?std;??

using?namespace?cv;??

int?CropImageCount=0;//裁剪出來的人體圖片個數??

/**?

*?通過根節點和節點名查找所有指定節點，結果放到節點數組NodeVector中?

*?@param?pRootEle?xml文件的根節點?

*?@param?strNodeName?要查詢的節點名?

*?@param?NodeVector?查詢到的節點指針數組?

*?@return?找到至少一個相應節點，返回true；否則false?

*/??

bool?GetAllNodePointerByName(TiXmlElement*?pRootEle,?string?strNodeName,?vector<TiXmlElement*>?&NodeVector)??

{??

????//如果NodeName等于根節點名，加入NodeVector數組??

????if(strNodeName?==?pRootEle->Value())??

????{??

????????NodeVector.push_back(pRootEle);//添加到數組末尾??

????????//這里根據VOC?Annotation的XML文件格式，認為相同節點名的節點不會有父子關系，所以所有相同節點名的節點都在同一級別上??

????????//只要找到第一個，剩下的肯定在它的兄弟節點里面??

????????for(TiXmlElement?*?pElement?=?pRootEle->NextSiblingElement();?pElement;?pElement?=?pElement->NextSiblingElement())??

????????????if(strNodeName?==?pElement->Value())??

????????????????NodeVector.push_back(pElement);??

????????return?true;??

????}??

????TiXmlElement?*?pEle?=?pRootEle;??

????for(pEle?=?pRootEle->FirstChildElement();?pEle;?pEle?=?pEle->NextSiblingElement())??

????{??

????????//遞歸處理子節點，獲取節點指針??

????????if(GetAllNodePointerByName(pEle,strNodeName,NodeVector))??

????????????return?true;??

????}??

????return?false;//沒找到??

}??

/**?

*?根據目標名過濾目標節點數組,刪除所有目標名不是objectName的元素?

*?@param?NodeVector?要操作的TiXmlElement元素指針數組?

*?@param?objectName?指定的目標名，刪除所有目標名不是objectName的元素?

*?@return?過濾后目標數組為空，返回false；否則返回true?

*/??

bool?FiltObject(vector<TiXmlElement*>?&NodeVector,?string?objectName)??

{??

????TiXmlElement?*?pEle?=?NULL;??

????vector<TiXmlElement?*>::iterator?iter?=?NodeVector.begin();//數組的迭代器??

????for(;?iter?!=?NodeVector.end();)??

????{??

????????pEle?=?*?iter;//第i個元素??

????????//若目標名不是objectName，刪除此節點??

????????if(?objectName?!=?pEle->FirstChildElement()->GetText()?)??

????????{??

????????????//cout<<"刪除的目標節點："<<pEle->FirstChildElement()->GetText()?<<endl;??

????????????iter?=?NodeVector.erase(iter);//刪除目標名不是objectName的，返回下一個元素的指針??

????????}??

????????else??

????????????iter++;??

????}??

????if(?0?==?NodeVector.size())//過濾后目標數組為空，說明不包含指定目標??

????????return?false;??

????else???

????????return?true;??

}??

/**?

*?根據每個目標的BoundingBox，剪裁圖像，保存為文件?

*?@param?img?圖像?

*?@param?NodeVector?目標節點數組?

*/??

void?CropImage(Mat?img,?vector<TiXmlElement*>?NodeVector)??

{??

????int?xmin,ymin,xmax,ymax;//從目標節點中讀出的包圍盒參數??

????char?fileName[256];//剪裁后的圖片和其水平翻轉圖片的文件名??

????//遍歷目標數組??

????vector<TiXmlElement?*>::iterator?iter?=?NodeVector.begin();//數組的迭代器??

????for(;?iter?!=?NodeVector.end();?iter++)??

????{??

????????//遍歷每個目標的子節點??

????????TiXmlElement?*pEle?=?(*iter)->FirstChildElement();//第i個元素的第一個孩子??

????????for(;?pEle;?pEle?=?pEle->NextSiblingElement())??

????????{??

????????????//找到包圍盒"bndbox"節點??

????????????if(string("bndbox")?==?pEle->Value())??

????????????{??

????????????????TiXmlElement?*?pCoord=?pEle->FirstChildElement();//包圍盒的第一個坐標值??

????????????????//依次遍歷包圍盒的4個坐標值，放入整型變量中??

????????????????for(;?pCoord;?pCoord?=?pCoord->NextSiblingElement())??

????????????????{??

????????????????????if(string("xmin")?==?pCoord->Value())??

????????????????????????xmin?=?atoi(pCoord->GetText());//xmin??

????????????????????if(string("ymin")?==?pCoord->Value())??

????????????????????????ymin?=?atoi(pCoord->GetText());//ymin??

????????????????????if(string("xmax")?==?pCoord->Value())??

????????????????????????xmax?=?atoi(pCoord->GetText());//xmax??

????????????????????if(string("ymax")?==?pCoord->Value())??

????????????????????????ymax?=?atoi(pCoord->GetText());//ymax??

????????????????}??

????????????????//cout<<"xmin:"<<xmin<<","<<"ymin:"<<ymin<<","<<"xmax:"<<xmax<<","<<"ymax:"<<ymax<<endl;;??

????????????????//根據讀取的包圍盒坐標設置圖像ROI??

????????????????Mat?imgROI?=?img(Rect(xmin,ymin,xmax-xmin,ymax-ymin));??

????????????????resize(imgROI,imgROI,Size(64,128));//縮放為64*128大小??

????????????????sprintf(fileName,"person%06d.jpg",++CropImageCount);//生成剪裁圖片的文件名??

????????????????imwrite(fileName,imgROI);//保存文件??

????????????????flip(imgROI,imgROI,1);//水平翻轉??

????????????????memset(fileName,0x00,sizeof(fileName));??

????????????????sprintf(fileName,"person%06d.jpg",++CropImageCount);//生成剪裁圖片的水平翻轉圖片的文件名??

????????????????imwrite(fileName,imgROI);//保存文件??

????????????}??

????????}??

????}??

}??

/**?

*?根據XML文件，從圖像中剪裁出objectName目標?

*?@param?XMLFile?XML文件名?

*?@param?img?對應的圖像?

*?@param?objectName?目標名?

*?@return?若圖像中包含objectName目標，返回true；否則返回false?

*/??

bool?CropImageAccordingToXML(string?XMLFile,?Mat?img,?string?objectName)??

{??

????TiXmlDocument?*?pDoc?=?new?TiXmlDocument();//創建XML文檔??

????pDoc->LoadFile(XMLFile.c_str());//裝載XML文件??

????vector<TiXmlElement*>?nodeVector;//節點數組??

????//查找所有節點名是object的節點，即目標節點，結果放到節點數組nodeVector中??

????if(?false?==?GetAllNodePointerByName(pDoc->RootElement(),?"object",?nodeVector)?)//未找到指定目標??

????????return?false;??

????//cout<<"所有目標個數："<<nodeVector.size()<<endl;??

????//過濾節點數組，刪除所有節點名不是objectName的節點??

????if(?false?==?FiltObject(nodeVector,objectName)?)//目標數組中沒有指定目標??

????????return?false;??

????//cout<<"過濾后的目標個數："<<nodeVector.size()<<endl;??

????//根據每個目標的BoundingBox，剪裁圖像，保存為文件??

????CropImage(img,nodeVector);??

}??

int?main()??

{??

????int?fileCount=0;//文件個數??

????Mat?src;??

????string?XMLName,ImgName;//XML文件名和對應的圖片文件名??

????//ifstream?fin("VOC2012AnnotationsXMLList.txt");//打開XML文件列表??

????ifstream?fin("subset.txt");??

????//ifstream?fin("test.txt");??

????//讀取XML文件列表??

????while(getline(fin,XMLName))??

????{??

????????cout<<"處理："<<XMLName<<endl;??

????????ImgName?=?"D:\\DataSet\\VOCtrainval_11-May-2012\\VOCdevkit\\VOC2012\\JPEGImages\\"?+?XMLName?+?".jpg";??

????????XMLName?=?"D:\\DataSet\\VOCtrainval_11-May-2012\\VOCdevkit\\VOC2012\\Annotations\\"?+?XMLName?+?".xml";??

????????src?=?imread(ImgName);??

????????CropImageAccordingToXML(XMLName,src,"person");//根據XML標注文檔，從圖像src中剪裁出所有person目標，保存為文件??

????}??

????system("pause");??

}??

源碼下載，環境為VS2010 + OpenCV2.4.4 + TinyXML2.6.2

http://download.csdn.net/detail/masikkk/6547823

編譯好的TinyXML2.6.2：

http://download.csdn.net/detail/masikkk/6547809

總結

以上是生活随笔為你收集整理的利用TinyXML读取VOC2012数据集的XML标注文件裁剪出所有人体目标保存为文件的全部內容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。

上一篇： Caffe代码阅读
下一篇： WPF messagebox生命周期的探