private void indexFiles(ArrayList<String> images, DocumentBuilder builder, String indexPath)
throws IOException {
// eventually check if the directory is there or not ...
IndexWriter iw = LuceneUtils.createIndexWriter(testIndex, false);
int count = 0;
long time = System.currentTimeMillis();
for (String identifier : images) {
// TODO: cut toes from the image ... -> doesn't work out very well. Stable at first,
// decreasing then.
// TODO: Joint Histogram ...
// TODO: LSA / PCA on the vectors ...-> this looks like a job for me :-D
// TODO: local features ...
Document doc = null;
if (cutImages) {
BufferedImage bimg =
ImageUtils.cropImage(ImageIO.read(new FileInputStream(identifier)), 0, 0, 200, 69);
doc = builder.createDocument(bimg, identifier);
} else doc = builder.createDocument(new FileInputStream(identifier), identifier);
iw.addDocument(doc);
count++;
if (count % 100 == 0) {
int percent = (int) Math.floor(((double) count * 100.0) / (double) images.size());
double timeTemp = (double) (System.currentTimeMillis() - time) / 1000d;
int secsLeft =
(int) Math.round(((timeTemp / (double) count) * (double) images.size()) - timeTemp);
System.out.println(percent + "% finished (" + count + " files), " + secsLeft + " s left");
}
}
long timeTaken = (System.currentTimeMillis() - time);
float sec = ((float) timeTaken) / 1000f;
System.out.println(sec + " seconds taken, " + (timeTaken / count) + " ms per image.");
iw.commit();
iw.close();
}
public double testIndexing() throws IOException, IllegalAccessException, InstantiationException {
LocalitySensitiveHashing.generateHashFunctions();
LocalitySensitiveHashing.readHashFunctions();
DocumentBuilder builder = new ChainedDocumentBuilder();
((ChainedDocumentBuilder) builder).addBuilder(DocumentBuilderFactory.getCEDDDocumentBuilder());
// System.out.println("-< Getting files to index >--------------");
ArrayList<String> images = FileUtils.getAllImages(new File(testExtensive), true);
// System.out.println("-< Indexing " + images.size() + " files >--------------");
IndexWriter iw =
LuceneUtils.createIndexWriter(indexPath, true, LuceneUtils.AnalyzerType.WhitespaceAnalyzer);
int count = 0;
long time = System.currentTimeMillis();
for (String identifier : images) {
CEDD cedd = new CEDD();
cedd.extract(ImageIO.read(new FileInputStream(identifier)));
Document doc = new Document();
doc.add(new Field(DocumentBuilder.FIELD_NAME_CEDD, cedd.getByteArrayRepresentation()));
doc.add(
new Field(
DocumentBuilder.FIELD_NAME_IDENTIFIER,
identifier,
Field.Store.YES,
Field.Index.NOT_ANALYZED));
int[] hashes = LocalitySensitiveHashing.generateHashes(cedd.getDoubleHistogram());
StringBuilder hash = new StringBuilder(512);
for (int i = 0; i < hashes.length; i++) {
hash.append(hashes[i]);
hash.append(' ');
}
// System.out.println("hash = " + hash);
doc.add(new Field("hash", hash.toString(), Field.Store.YES, Field.Index.ANALYZED));
iw.addDocument(doc);
count++;
// if (count % 100 == 0) System.out.println(count + " files indexed.");
}
long timeTaken = (System.currentTimeMillis() - time);
float sec = ((float) timeTaken) / 1000f;
// System.out.println(sec + " seconds taken, " + (timeTaken / count) + " ms per image.");
iw.close();
return testSearch();
}
private void indexFiles(ArrayList<String> images, DocumentBuilder builder, String indexPath)
throws IOException {
// System.out.println(">> Indexing " + images.size() + " files.");
// DocumentBuilder builder = DocumentBuilderFactory.getExtensiveDocumentBuilder();
// DocumentBuilder builder = DocumentBuilderFactory.getFastDocumentBuilder();
IndexWriter iw = LuceneUtils.createIndexWriter(indexPath, true);
int count = 0;
long time = System.currentTimeMillis();
for (String identifier : images) {
Document doc = builder.createDocument(new FileInputStream(identifier), identifier);
iw.addDocument(doc);
count++;
if (count % 100 == 0) System.out.println(count + " files indexed.");
// if (count == 200) break;
}
long timeTaken = (System.currentTimeMillis() - time);
float sec = ((float) timeTaken) / 1000f;
System.out.println(sec + " seconds taken, " + (timeTaken / count) + " ms per image.");
iw.commit();
iw.close();
}
protected void setUp() throws Exception {
super.setUp();
// set to all queries ... approach "leave one out"
sampleQueries = new int[1000];
for (int i = 0; i < sampleQueries.length; i++) {
sampleQueries[i] = i;
}
indexPath += "-" + System.currentTimeMillis() % (1000 * 60 * 60 * 24 * 7);
// Setting up DocumentBuilder:
// parallelIndexer = new ParallelIndexer(8, indexPath, testExtensive);
parallelIndexer =
new ParallelIndexer(8, indexPath, testExtensive, true) {
@Override
public void addBuilders(ChainedDocumentBuilder builder) {
// builder.addBuilder(DocumentBuilderFactory.getCEDDDocumentBuilder());
//
// builder.addBuilder(DocumentBuilderFactory.getAutoColorCorrelogramDocumentBuilder());
builder.addBuilder(DocumentBuilderFactory.getColorLayoutBuilder());
builder.addBuilder(DocumentBuilderFactory.getEdgeHistogramBuilder());
// builder.addBuilder(DocumentBuilderFactory.getFCTHDocumentBuilder());
builder.addBuilder(DocumentBuilderFactory.getJCDDocumentBuilder());
//
// builder.addBuilder(DocumentBuilderFactory.getJointHistogramDocumentBuilder());
builder.addBuilder(DocumentBuilderFactory.getOpponentHistogramDocumentBuilder());
builder.addBuilder(DocumentBuilderFactory.getPHOGDocumentBuilder());
//
// builder.addBuilder(DocumentBuilderFactory.getColorHistogramDocumentBuilder());
builder.addBuilder(DocumentBuilderFactory.getScalableColorBuilder());
//
// builder.addBuilder(DocumentBuilderFactory.getLuminanceLayoutDocumentBuilder());
//
// builder.addBuilder(DocumentBuilderFactory.getJpegCoefficientHistogramDocumentBuilder());
//
// builder.addBuilder(DocumentBuilderFactory.getColorHistogramDocumentBuilder());
// builder.addBuilder(DocumentBuilderFactory.getGaborDocumentBuilder());
// builder.addBuilder(DocumentBuilderFactory.getTamuraDocumentBuilder());
// builder.addBuilder(DocumentBuilderFactory.getScalableColorBuilder());
// builder.addBuilder(new GenericDocumentBuilder(RankAndOpponent.class,
// "jop"));
// builder.addBuilder(new
// GenericFastDocumentBuilder(FuzzyOpponentHistogram.class, "opHist"));
// builder.addBuilder(new SurfDocumentBuilder());
// builder.addBuilder(new MSERDocumentBuilder());
// builder.addBuilder(new SiftDocumentBuilder());
// builder.addBuilder(new GenericDocumentBuilder(SPCEDD.class,
// "spcedd"));
// builder.addBuilder(new GenericDocumentBuilder(SPFCTH.class,
// "spfcth"));
// builder.addBuilder(new GenericDocumentBuilder(SPJCD.class, "spjcd"));
// builder.addBuilder(new GenericDocumentBuilder(SPACC.class, "spacc"));
// builder.addBuilder(new
// GenericDocumentBuilder(LocalBinaryPatterns.class, "lbp"));
// builder.addBuilder(new
// GenericDocumentBuilder(BinaryPatternsPyramid.class, "whog"));
// builder.addBuilder(new
// GenericDocumentBuilder(LocalBinaryPatternsAndOpponent.class, "jhl"));
// builder.addBuilder(new
// GenericDocumentBuilder(RotationInvariantLocalBinaryPatterns.class, "rlbp"));
// builder.addBuilder(new GenericDocumentBuilder(SPLBP.class, "splbp"));
}
};
}
public void computeMAP(ImageSearcher searcher, String prefix, IndexReader reader)
throws IOException {
Pattern p = Pattern.compile("([0-9]+).jpg");
double map = 0;
double errorRate = 0d;
double precision10 = 0d;
double[] pr10cat = new double[10];
double[] pr10cnt = new double[10];
for (int i = 0; i < pr10cat.length; i++) {
pr10cat[i] = 0d;
pr10cnt[i] = 0d;
}
long sum = 0, ms = 0;
for (int i = 0; i < sampleQueries.length; i++) {
int id = sampleQueries[i];
String file = testExtensive + "/" + id + ".jpg";
ms = System.currentTimeMillis();
ImageSearchHits hits = searcher.search(findDoc(reader, id + ".jpg"), reader);
sum += (System.currentTimeMillis() - ms);
int goodOnes = 0;
double avgPrecision = 0d;
double precision10temp = 0d;
int countResults = 0;
for (int j = 0; j < hits.length(); j++) {
Document d = hits.doc(j);
String hitsId = d.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0];
Matcher matcher = p.matcher(hitsId);
if (matcher.find()) hitsId = matcher.group(1);
else fail("Did not get the number ...");
int testID = Integer.parseInt(hitsId);
if (testID != id) countResults++;
if ((testID != id) && ((int) Math.floor(id / 100) == (int) Math.floor(testID / 100))) {
goodOnes++;
// Only if there is a change in recall
avgPrecision += (double) goodOnes / (double) countResults;
// System.out.print("x");
if (j <= 10) {
precision10temp += 1d;
}
} else {
if (j == 1) { // error rate
errorRate++;
}
}
} // end for loop iterating results.
// if (avgPrecision<=0) {
// System.out.println("avgPrecision = " + avgPrecision);
// System.out.println("goodOnes = " + goodOnes);
// }
assertTrue("Check if average precision is > 0", avgPrecision > 0);
assertTrue("Check if goodOnes is > 0", goodOnes > 0);
avgPrecision = avgPrecision / goodOnes;
precision10 += precision10temp / 10d;
// precision @ 10 for each category ...
pr10cat[(int) Math.floor(id / 100)] += precision10temp / 10d;
pr10cnt[(int) Math.floor(id / 100)] += 1d;
map += avgPrecision;
}
map = map / sampleQueries.length;
errorRate = errorRate / sampleQueries.length;
precision10 = precision10 / sampleQueries.length;
System.out.print(prefix + "\t");
System.out.print(String.format("%.5f", map) + '\t');
System.out.print(String.format("%.5f", precision10) + '\t');
System.out.print(String.format("%.5f", errorRate) + '\t');
// precision@10 per category
for (int i = 0; i < pr10cat.length; i++) {
double v = 0;
if (pr10cnt[i] > 0) v = pr10cat[i] / pr10cnt[i];
// System.out.print(i + ": ");
System.out.printf("%.5f\t", v);
}
System.out.printf("%2.3f\t", (double) sum / (double) sampleQueries.length);
System.out.println();
}