haarlearner.cpp

MultiBoost 是c++实现的多类adaboost酸法。与传统的adaboost算法主要解决二类分类问题不同

/*
* This file is part of MultiBoost, a multi-class 
* AdaBoost learner/classifier
*
* Copyright (C) 2005-2006 Norman Casagrande
* For informations write to [e-mail]
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
*
*/

#include "HaarLearner.h"
#include "Others/HaarFeatures.h"
#include "Utils/StreamTokenizer.h"

#include "IO/HaarData.h"
#include "IO/Serialization.h"
#include "Utils/Args.h"

#include <limits> // for numeric_limits
#include <ctime> // for time

namespace MultiBoost {

// ------------------------------------------------------------------------------

void HaarLearner::declareArguments(nor_utils::Args& args)
{
   args.declareArgument("ftypes", 
                        "Specify the list with the type of Haar-like features. This list can contain "
                        "all the types requested at once. The types allowed are:\n"
                        "  2v: 2 blocs vertical feature\n"
                        "  2h: 2 blocs horizontal feature\n"
                        "  3v: 3 blocs vertical feature\n"
                        "  3h: 3 blocs horizontal feature\n"
                        "  4q: 4 blocs squared feature\n"
                        "For instance if you need all the 2 and 3 blocs type, just write:\n"
                        "-ftypes 2v2h3v3h\n"
                        "(DEFAULT: ALL of them will be selected)"
                        , 1, "<list>");

   args.declareArgument("csample",
                        "Instead of computing all the possible combinations, select a set of "
                        "random configurations. It is possible to specify the number of combinations "
                        "or the amount of time per feature type. The options are:\n"
                        "  num: The number of combination per type per iteration.\n"
                        "  time: The amount of time (in seconds) per type per iteration.\n"
                        "Example: -csample num 4 -> Set the number of random configurations to 4.\n"
                        "Example: -csample time 120 -> Set the time to 2 minutes per type per iteration.",
                        2, "<opt> <#|sec>");

   args.declareArgument("seed", "Defines the seed for the random sampling.", 1, "<seedval>");

   args.declareArgument("iisize", 
                        "Specify the size of the integral image representation for each example,\n"
                        "Example: -iisize 128x64", 
                        1, "<width>x<height>");

}

// ------------------------------------------------------------------------------

void HaarLearner::initOptions(nor_utils::Args& args)
{
   // If the sampling is required
   if (args.hasArgument("csample"))
   {
      string randSearch = args.getValue<string>("csample", 0);
      // if the type is ST_NUM this is the number of samplings per feature per iteration
      // if the type is ST_TIME this is the amount of time per feature per iteration
      _samplingVal = args.getValue<int>("csample", 1);

      if (randSearch == "num")
         _samplingType = ST_NUM;
      else if (randSearch == "time")
         _samplingType = ST_TIME;
   }

   // defines the seed of the sampling method.
   if (args.hasArgument("seed"))
   {
      unsigned int seed = args.getValue<unsigned int>("seed", 0);
      srand(seed);
   }

}

// ------------------------------------------------------------------------------

void HaarLearner::save(ofstream& outputStream, int numTabs)
{
   // save the type of feature
   outputStream << Serialization::standardTag("type", _pSelectedFeature->getShortName(), numTabs) << endl;

   // Save the coordinates and size of the feature
   outputStream << Serialization::getTabs(numTabs) << "<rect>" << endl;
      outputStream << Serialization::standardTag("x", _selectedConfig.x, numTabs+1) << endl;
      outputStream << Serialization::standardTag("y", _selectedConfig.y, numTabs+1) << endl;
      outputStream << Serialization::standardTag("width", _selectedConfig.width, numTabs+1) << endl;
      outputStream << Serialization::standardTag("height", _selectedConfig.height, numTabs+1) << endl;
   outputStream << Serialization::getTabs(numTabs) << "</rect>" << endl;

}

// -----------------------------------------------------------------------

void HaarLearner::load(nor_utils::StreamTokenizer& st)
{
   // get the type of feature
   string shortName = UnSerialization::seekAndParseEnclosedValue<string>(st, "type");
   // get it from the registered features
   _pSelectedFeature = HaarFeature::RegisteredFeatures().getFeature(shortName);

   // Load the coordinates and size of the feature
   _selectedConfig.x = UnSerialization::seekAndParseEnclosedValue<short>(st, "x");
   _selectedConfig.y = UnSerialization::seekAndParseEnclosedValue<short>(st, "y");
   _selectedConfig.width = UnSerialization::seekAndParseEnclosedValue<short>(st, "width");
   _selectedConfig.height = UnSerialization::seekAndParseEnclosedValue<short>(st, "height");

}

// ------------------------------------------------------------------------------

} // end of MultiBoost namespace