ANN.zip_ANN opencv _ANN_MLP_MLP_opencv ann_opencv mlp

/////*----------------------------------------------------------------------------- //// * Project.....................: Visual Computing at MICA //// * Copyright...................: (C) 2009 MICA. All rights reserved. //// * This project developed by...: MICA, HUT, Viet Nam http://mica.edu.vn //// * File........................: //// * Authors.....................: Tran Thi Thanh Hai //// * Description.................: Skin Detection Under Changing illumination using Neural Network //// * Creation....................: 22.01.2010 //// * Revision....................: //// -------------------------------------------------------------------------------*/ //// ////standard includes #include <stdlib.h> #include <stdio.h> #include <iostream> #include <fstream> #include <string> #include <cstdlib> #include <cstdio> #include <iostream> #include <iomanip> #include <sstream> #include <vector> #include "ml.h" #include <stdio.h> // This function reads data and responses from the file <filename> #include <stdlib.h> #include <stdio.h> #include <iostream> #include <string> #include <cstdlib> #include <cstdio> #include <iostream> #include <iomanip> #include <sstream> #include <vector> #include <fstream> //open CV includes #include <ml.h> using namespace std; ofstream outPre ("OutPre.txt"); CvANN_MLP machineBrain; static int build_mlp_classifier() { int var_count; CvMat* datas; CvMat* responses; // ofstream outFile ("OutTrain.txt"); // if (outFile.is_open()) //{ // ifstream inFile; // inFile.open("Train.txt"); // // if (!inFile) { // cerr << "Unable to open file Train.txt"; // exit(1); // call system to stop // } // // inFile >> var_count; // cout << "number of data pairs = " << var_count << endl; // cout << "Processing..."; // outFile << "number of data pairs = " << var_count << "\n"; // datas = cvCreateMat(var_count, 3, CV_32FC1); // responses = cvCreateMat(var_count, 3, CV_32FC1); // // int id = 0; // while (!inFile.eof()) // { // for (int i = 0; i < 3; i++) // { // int x ; // inFile >> x; // //cout << x << " " ; // //outFile << x << " "; // cvmSet(datas,id, i, x); // } // for (int i = 0; i < 3; i++) // { // int x ; // inFile >> x; // //cout << x << " " ; // //outFile << x << " "; // cvmSet(responses,id, i, x); // } // //cout << endl; // //outFile << "\n"; // id++; // } // inFile.close(); // } // ifstream inFile; inFile.open("Train.txt"); if (!inFile) { cerr << "Unable to open file Train.txt"; exit(1); // call system to stop } inFile >> var_count; cout << "number of data pairs = " << var_count << endl; cout << "Processing..."; datas = cvCreateMat(var_count, 3, CV_32FC1); responses = cvCreateMat(var_count, 3, CV_32FC1); int id = 0; while (!inFile.eof()) { for (int i = 0; i < 3; i++) { int x ; inFile >> x; //cout << x; cvmSet(datas,id, i, x); } for (int i = 0; i < 3; i++) { int x ; inFile >> x; cvmSet(responses,id, i, x); } id++; } inFile.close(); //The weight of each training data sample. We'll later set all to equal weights. CvMat* sampleWts = cvCreateMat(var_count, 1, CV_32FC1); //The matrix representation of our ANN. We'll have four layers. CvMat* neuralLayers = cvCreateMat(3, 1, CV_32SC1); CvMat neuralLayers1, sampleWts1; cvGetRows(sampleWts, &sampleWts1, 0, var_count); cvGetRows(neuralLayers, &neuralLayers1, 0, 3); // 0,3 //Setting the number of neurons on each layer of the ANN /* We have in Layer 1: 3 neurons (3 inputs) Layer 2: 20 neurons (hidden layer) Layer 3: 3 neurons (3 output) */ cvSet1D(&neuralLayers1, 0, cvScalar(3)); cvSet1D(&neuralLayers1, 1, cvScalar(20)); cvSet1D(&neuralLayers1, 2, cvScalar(3)); for (int i=0; i<var_count; i++) { cvSet1D(&sampleWts1, i, cvScalar(1)); } //Create our ANN. machineBrain.create(neuralLayers); //Train it with our data. //See the Machine learning reference at http://www.seas.upenn.edu/~bensapp/opencvdocs/ref/opencvref_ml.htm#ch_ann int iters = machineBrain.train( datas, responses, sampleWts, 0, CvANN_MLP_TrainParams( cvTermCriteria( CV_TERMCRIT_ITER+CV_TERMCRIT_EPS, 100000, 1.0 ),// CvANN_MLP_TrainParams::BACKPROP, 0.01, 0.05 ) ); /*outFile << "number of trainning iterations = " << iters << "\n"; outFile << "end of trainning" << "\n"; outFile.close();*/ return iters; return 0; } // Predict the output with the trained ANN given the three inputs. void Predict(float data1, float data2, float data3) { float _sample[3]; CvMat sample = cvMat(1, 3, CV_32FC1, _sample); float _predout[3]; CvMat predout = cvMat(1, 3, CV_32FC1, _predout); sample.data.fl[0] = data1; sample.data.fl[1] = data2; sample.data.fl[2] = data3; machineBrain.predict(&sample, &predout); //cout<<min((int)predout.data.fl[0],255)<<" "<<min((int)predout.data.fl[1],255)<<" "<<min((int)predout.data.fl[2],255)<<endl; outPre<<min((int)predout.data.fl[0],255)<<" "<<min((int)predout.data.fl[1],255)<<" "<<min((int)predout.data.fl[2],255)<<" "; } int main( int argc, char *argv[] ) { int iters = build_mlp_classifier(); //for (int i=0; i<5; i++) build_mlp_classifier(); //cout << "number of trainning iterations = " << iters << endl; //cout << "end of trainning, please enter a key to exit" << endl; //cout << endl; ifstream inFile; inFile.open("TestImage.txt"); if (outPre.is_open()) { while (!inFile.eof()) { float data1, data2, data3 ; inFile >> data1; inFile >> data2; inFile >> data3; //cout <<(int)data1<<" "<<(int)data2<<" "<<(int) data3<< endl; //if ((data1 + data2 + data3)/3 > 10) Predict(data1,data2,data3); } inFile.close(); } outPre.close(); cout << "Complete!"; int wait; scanf("%d",&wait); return 0; }