/*
* This file is part of the LIRE project: http://www.semanticmetadata.net/lire
* LIRE is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* LIRE 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 General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with LIRE; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*
* We kindly ask you to refer the any or one of the following publications in
* any publication mentioning or employing Lire:
*
* Lux Mathias, Savvas A. Chatzichristofis. Lire: Lucene Image Retrieval –
* An Extensible Java CBIR Library. In proceedings of the 16th ACM International
* Conference on Multimedia, pp. 1085-1088, Vancouver, Canada, 2008
* URL: http://doi.acm.org/10.1145/1459359.1459577
*
* Lux Mathias. Content Based Image Retrieval with LIRE. In proceedings of the
* 19th ACM International Conference on Multimedia, pp. 735-738, Scottsdale,
* Arizona, USA, 2011
* URL: http://dl.acm.org/citation.cfm?id=2072432
*
* Mathias Lux, Oge Marques. Visual Information Retrieval using Java and LIRE
* Morgan & Claypool, 2013
* URL: http://www.morganclaypool.com/doi/abs/10.2200/S00468ED1V01Y201301ICR025
*
* Copyright statement:
* ====================
* (c) 2002-2013 by Mathias Lux (mathias@juggle.at)
* http://www.semanticmetadata.net/lire, http://www.lire-project.net
*
* Updated: 26.08.14 12:56
*/
package net.semanticmetadata.lire.benchmarking;
import junit.framework.TestCase;
import net.semanticmetadata.lire.DocumentBuilder;
import net.semanticmetadata.lire.ImageSearchHits;
import net.semanticmetadata.lire.ImageSearcher;
import net.semanticmetadata.lire.impl.BitSamplingImageSearcher;
import net.semanticmetadata.lire.impl.GenericFastImageSearcher;
import net.semanticmetadata.lire.impl.SimpleImageSearchHits;
import net.semanticmetadata.lire.indexing.hashing.BitSampling;
import net.semanticmetadata.lire.indexing.hashing.LocalitySensitiveHashing;
import net.semanticmetadata.lire.indexing.tools.ParallelExtractor;
import net.semanticmetadata.lire.indexing.tools.ProximityHashingIndexor;
import org.apache.lucene.index.DirectoryReader;
import org.apache.lucene.index.IndexReader;
import org.apache.lucene.store.FSDirectory;
import org.apache.lucene.store.IOContext;
import org.apache.lucene.store.RAMDirectory;
import java.io.File;
import java.io.FileInputStream;
import java.io.IOException;
import java.util.ArrayList;
import java.util.HashSet;
import java.util.Iterator;
/**
* Benchmarking class for hashing based indexes.
*
* @author Mathias Lux, mathias@juggle.at, 01.06.13
*/
public class TestHashingIndex extends TestCase {
private String dataSetImageList = "D:\\Java\\Projects\\Lire\\testdata\\flickrphotos.lst";
// private String dataSetImageList = "D:\\DataSets\\Flickrphotos\\imageList.txt";
private String dataSetDataOut = "D:\\Java\\Projects\\Lire\\testdata\\flickrphotos.out";
// private String dataSetDataOut = "D:\\DataSets\\Flickrphotos\\imageList.out";
private String testIndex = "C:/Temp/idx-test-hashing";
public void testExtractFeatures() {
ParallelExtractor pe = new ParallelExtractor();
pe.setFileList(new File(dataSetImageList));
pe.setOutFile(new File(dataSetDataOut));
ParallelExtractor.setNumberOfThreads(6);
pe.addFeature(new PHOG());
pe.addFeature(new CEDD());
// pe.addFeature(new JCD());
// pe.addFeature(new FCTH());
pe.addFeature(new AutoColorCorrelogram());
pe.addFeature(new OpponentHistogram());
pe.addFeature(new SimpleColorHistogram());
pe.addFeature(new ColorLayout());
pe.addFeature(new EdgeHistogram());
// pe.addFeature(new SPCEDD());
pe.run();
}
public void testHashing() throws IOException, IllegalAccessException, InstantiationException {
BitSampling.setBits(12);
BitSampling.setNumFunctionBundles(150);
// testHashing(JCD.class, DocumentBuilder.FIELD_NAME_JCD);
testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// testHashing(FCTH.class, DocumentBuilder.FIELD_NAME_FCTH);
testHashing(OpponentHistogram.class, DocumentBuilder.FIELD_NAME_OPPONENT_HISTOGRAM);
testHashing(EdgeHistogram.class, DocumentBuilder.FIELD_NAME_EDGEHISTOGRAM);
testHashing(PHOG.class, DocumentBuilder.FIELD_NAME_PHOG);
testHashing(AutoColorCorrelogram.class, DocumentBuilder.FIELD_NAME_AUTOCOLORCORRELOGRAM);
testHashing(SimpleColorHistogram.class, DocumentBuilder.FIELD_NAME_COLORHISTOGRAM);
testHashing(ColorLayout.class, DocumentBuilder.FIELD_NAME_COLORLAYOUT);
// testHashing(SPCEDD.class, "spcedd");
// BitSampling.bits = 12;
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// BitSampling.bits = 16;
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// BitSampling.bits = 24;
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// BitSampling.bits = 28;
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
//
// BitSampling.setNumFunctionBundles(50);
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// BitSampling.setNumFunctionBundles(150);
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// BitSampling.setNumFunctionBundles(200);
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// BitSampling.setNumFunctionBundles(250);
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
//
// BitSampling.bits = 12;
// BitSampling.setNumFunctionBundles(150);
// BitSampling.setW(0.01d);
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// testHashing(AutoColorCorrelogram.class, DocumentBuilder.FIELD_NAME_AUTOCOLORCORRELOGRAM);
// BitSampling.setW(0.1d);
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// testHashing(AutoColorCorrelogram.class, DocumentBuilder.FIELD_NAME_AUTOCOLORCORRELOGRAM);
// BitSampling.setW(0.5d);
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// testHashing(AutoColorCorrelogram.class, DocumentBuilder.FIELD_NAME_AUTOCOLORCORRELOGRAM);
// BitSampling.setW(4d);
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// BitSampling.setW(5d);
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
// BitSampling.setW(10d);
// testHashing(CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
}
private void testHashing(Class featureClass, String fieldName) throws IOException, InstantiationException, IllegalAccessException {
String hashesFile = "hashes.obj";
String hashesFileL = "l_hashes.obj";
int numResults = 50;
int maxQueries = 20;
int queryOffset = 100;
File file = new File(hashesFile);
if (file.exists()) file.delete();
file = new File(hashesFileL);
if (file.exists()) file.delete();
BitSampling.generateHashFunctions(hashesFile);
LocalitySensitiveHashing.generateHashFunctions(hashesFileL);
// HashingIndexor hi = new HashingIndexor();
ProximityHashingIndexor hi = new ProximityHashingIndexor();
BitSampling.readHashFunctions(new FileInputStream(hashesFile));
LocalitySensitiveHashing.readHashFunctions(new FileInputStream(hashesFileL));
hi.setFeatureClass(featureClass);
hi.addInputFile(new File(dataSetDataOut));
hi.setIndexPath(testIndex);
hi.run();
System.out.println();
IndexReader reader = DirectoryReader.open(new RAMDirectory(FSDirectory.open(new File(testIndex)), IOContext.READONCE));
// generating ground truth for all queries ...
ImageSearcher groundTruth = new GenericFastImageSearcher(numResults, featureClass, fieldName);
ArrayList<ImageSearchHits> trueHitsList = new ArrayList<ImageSearchHits>(maxQueries);
long time = System.currentTimeMillis();
for (int q = 0; q < maxQueries; q++) {
trueHitsList.add(q, groundTruth.search(reader.document(q + queryOffset), reader));
}
time = System.currentTimeMillis() - time;
// header
System.out.println(featureClass.getName().substring(featureClass.getName().lastIndexOf('.') + 1));
System.out.println("Number of queries: " + maxQueries);
System.out.println("Time taken for linear search: " + (time / maxQueries));
System.out.printf("numFunctionBundles: %d, numBits: %d, w: %2.2f, dimensions: %d\n", BitSampling.getNumFunctionBundles(), BitSampling.getBits(), BitSampling.getW(), BitSampling.dimensions);
System.out.println("#hashedResults\ttrue pos.\t#results\tms per search\tprecision");
for (int j = 100; j <= 3000; j += 100) {
ImageSearcher hashed = new BitSamplingImageSearcher(numResults, fieldName, fieldName + "_hash", (LireFeature) featureClass.newInstance(), new FileInputStream(hashesFile), j);
long ms = 0;
long msSum = 0;
int posSum = 0;
for (int q = 0; q < maxQueries; q++) {
ms = System.currentTimeMillis();
ImageSearchHits hashedHits = hashed.search(reader.document(q + queryOffset), reader);
assert(hashedHits.length()<=numResults);
msSum += System.currentTimeMillis() - ms;
HashSet<Integer> t = new HashSet<Integer>(hashedHits.length());
HashSet<Integer> h = new HashSet<Integer>(hashedHits.length());
for (int i = 0; i < trueHitsList.get(q).length(); i++) {
t.add(((SimpleImageSearchHits) trueHitsList.get(q)).readerID(i));
h.add(((SimpleImageSearchHits) hashedHits).readerID(i));
}
assert (t.size() == h.size());
int intersect = 0;
for (Iterator<Integer> iterator = h.iterator(); iterator.hasNext(); ) {
if (t.contains(iterator.next())) {
intersect++;
}
}
posSum += intersect;
}
if (j > 1400) j += 100;
double truePositives = ((double) posSum) / ((double) maxQueries);
System.out.printf("%4d\t%4.1f\t%4d\t%6.1f\t%1.3f\n", j, truePositives, numResults, ((double) msSum) / ((double) maxQueries), truePositives / (double) numResults);
if (posSum / maxQueries == numResults) break;
}
}
}