public void computeErrorRate(ImageSearcher searcher, String prefix)
throws IOException, InstantiationException, IllegalAccessException {
// int maxHits = 10;
IndexReader reader = DirectoryReader.open(FSDirectory.open(new File(testIndex)));
for (Iterator<String> testIterator = testcases.keySet().iterator(); testIterator.hasNext(); ) {
queryImage = testIterator.next();
Document query;
if (cutImages) {
BufferedImage bimg =
ImageUtils.cropImage(ImageIO.read(new FileInputStream(queryImage)), 0, 0, 200, 69);
query = builder.createDocument(new FileInputStream(queryImage), queryImage);
} else query = builder.createDocument(new FileInputStream(queryImage), queryImage);
ImageSearchHits hits = searcher.search(query, reader);
// hits = rerank(hits, query, ColorLayout.class, DocumentBuilder.FIELD_NAME_COLORLAYOUT);
for (int i = 0; i < hits.length(); i++) {
if (hits.doc(i)
.get("descriptorImageIdentifier")
.toLowerCase()
.endsWith(testcases.get(queryImage))) {
System.out.println(
queryImage.substring(queryImage.lastIndexOf('\\') + 1)
+ "-"
+ prefix
+ " -> Found at rank "
+ i
+ " ("
+ hits.length()
+ ")");
}
}
// saveToHtml(queryImage.substring(queryImage.lastIndexOf('\\') + 1) + "-" + prefix, hits,
// queryImage);
}
}
public void testParallelMAP() throws IOException {
int maxHits = 1000;
IndexReader reader = IndexReader.open(FSDirectory.open(new File(indexPath)));
ParallelImageSearcher searcher;
searcher = new ParallelImageSearcher(maxHits, CEDD.class, DocumentBuilder.FIELD_NAME_CEDD);
Pattern p = Pattern.compile("([0-9]+).jpg");
double map = 0;
double errorRate = 0d;
for (int i = 0; i < sampleQueries.length; i++) {
int id = sampleQueries[i];
// System.out.println("id = " + id + ": " + "("+i+")");
String file = testExtensive + "/" + id + ".jpg";
String[] files = {
id + ".jpg", (id + 1) + ".jpg", (id + 2) + ".jpg", (id + 3) + ".jpg", (id + 4) + ".jpg"
};
ImageSearchHits[] hits = searcher.search(findDocs(reader, files), reader);
for (int k = 0; k < hits.length; k++) {
int currentID = id + k;
ImageSearchHits h = hits[k];
int goodOnes = 0;
double avgPrecision = 0;
for (int j = 0; j < h.length(); j++) {
Document d = h.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 != currentID)
&& ((int) Math.floor(id / 100) == (int) Math.floor(testID / 100))) {
goodOnes++;
// Only if there is a change in recall
avgPrecision += (double) goodOnes / (double) (j + 1);
// System.out.print("x");
} else {
if (j == 1) { // error rate
errorRate++;
}
}
// System.out.print(" (" + testID + ") ");
}
assertTrue(goodOnes > 0);
avgPrecision = avgPrecision / goodOnes;
assertTrue(avgPrecision > 0);
map += avgPrecision;
// System.out.println(" " + avgPrecision + " (" + map / (i + 1) + ")");
}
}
assertTrue(sampleQueries.length > 0);
map = map / sampleQueries.length;
errorRate = errorRate / sampleQueries.length;
System.out.println("map = " + map);
System.out.println("errorRate = " + errorRate);
}
private ImageSearchHits rerank(
ImageSearchHits hits, Document query, Class descriptorClass, String fieldName)
throws IllegalAccessException, InstantiationException {
ArrayList<SimpleResult> results = new ArrayList<SimpleResult>(hits.length());
LireFeature qf = getFeature(descriptorClass, query.getValues(fieldName)[0]);
float maxDistance = 0f;
for (int i = 0; i < hits.length(); i++) {
LireFeature lf = getFeature(descriptorClass, hits.doc(i).getValues(fieldName)[0]);
float distance = lf.getDistance(qf);
SimpleResult sr = new SimpleResult(distance, hits.doc(i), i);
results.add(sr);
maxDistance = Math.max(maxDistance, distance);
}
Collections.sort(results);
return new SimpleImageSearchHits(results, maxDistance);
}
public void testSearch() throws IOException {
IndexReader reader = DirectoryReader.open(FSDirectory.open(new File(indexPath)));
int numDocs = reader.numDocs();
System.out.println("numDocs = " + numDocs);
ImageSearcher searcher = ImageSearcherFactory.createScalableColorImageSearcher(50);
FileInputStream imageStream = new FileInputStream(testFilesPath + testFiles[0]);
BufferedImage bimg = ImageIO.read(imageStream);
ImageSearchHits hits = null;
long time = System.currentTimeMillis();
for (int i = 0; i < numsearches; i++) {
hits = searcher.search(bimg, reader);
}
time = System.currentTimeMillis() - time;
System.out.println(
((float) time / (float) numsearches)
+ " ms per search with image, averaged on "
+ numsearches);
for (int i = 0; i < 5; i++) {
System.out.println(
hits.score(i)
+ ": "
+ hits.doc(i).getField(DocumentBuilder.FIELD_NAME_IDENTIFIER).stringValue());
}
Document document = hits.doc(0);
time = System.currentTimeMillis();
for (int i = 0; i < numsearches; i++) {
hits = searcher.search(document, reader);
}
time = System.currentTimeMillis() - time;
System.out.println(
((float) time / (float) numsearches)
+ " ms per search with document, averaged on "
+ numsearches);
for (int i = 0; i < 5; i++) {
System.out.println(
hits.score(i)
+ ": "
+ hits.doc(i).getField(DocumentBuilder.FIELD_NAME_IDENTIFIER).stringValue());
}
document = getDocumentBuilder().createDocument(bimg, testFilesPath + testFiles[0]);
time = System.currentTimeMillis();
for (int i = 0; i < numsearches; i++) {
hits = searcher.search(document, reader);
}
time = System.currentTimeMillis() - time;
System.out.println(
((float) time / (float) numsearches)
+ " ms per search with document, averaged on "
+ numsearches);
for (int i = 0; i < 5; i++) {
System.out.println(
hits.score(i)
+ ": "
+ hits.doc(i).getField(DocumentBuilder.FIELD_NAME_IDENTIFIER).stringValue());
}
}
public void tttestMAPLocalFeatureHistogram() throws IOException {
int maxSearches = 200;
int maxHits = 100;
IndexReader reader = IndexReader.open(FSDirectory.open(new File(indexPath)));
IndexSearcher is = new IndexSearcher(reader);
ImageSearcher searcher;
// searcher = new SiftLocalFeatureHistogramImageSearcher(maxHits);
searcher = ImageSearcherFactory.createColorHistogramImageSearcher(maxHits);
// searcher = ImageSearcherFactory.createCEDDImageSearcher(maxHits);
// searcher = ImageSearcherFactory.createFCTHImageSearcher(maxHits);
Pattern p = Pattern.compile("\\\\\\d+\\.jpg");
double map = 0;
for (int i = 0; i < sampleQueries.length; i++) {
int id = sampleQueries[i];
System.out.print("id = " + id + ": ");
String file = testExtensive + "/" + id + ".jpg";
ImageSearchHits hits = searcher.search(findDoc(reader, id + ".jpg"), reader);
int goodOnes = 0;
double avgPrecision = 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 = hitsId.substring(matcher.start() + 1, hitsId.lastIndexOf("."));
else fail("Did not get the number ...");
int testID = Integer.parseInt(hitsId);
// System.out.print(". " + hitsId + "/" +
// d.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0]+ " ");
if ((int) Math.floor(id / 100) == (int) Math.floor(testID / 100)) {
goodOnes++;
System.out.print("x");
} else {
System.out.print("o");
}
// System.out.print(" (" + testID + ") ");
avgPrecision += (double) goodOnes / (double) (j + 1);
}
avgPrecision = avgPrecision / hits.length();
map += avgPrecision;
System.out.println(" " + avgPrecision + " (" + map / (i + 1) + ")");
}
map = map / sampleQueries.length;
System.out.println("map = " + map);
}
private void saveToHtml(String prefix, ImageSearchHits hits, String queryImage)
throws IOException {
BufferedWriter bw = new BufferedWriter(new FileWriter("results - " + prefix + ".html"));
bw.write(
"<html>\n"
+ "<head><title>Search Results</title></head>\n"
+ "<body bgcolor=\"#FFFFFF\">\n");
bw.write("<h3>query</h3>\n");
bw.write(
"<a href=\"file://" + queryImage + "\"><img src=\"file://" + queryImage + "\"></a><p>\n");
bw.write("<h3>results</h3>\n");
for (int i = 0; i < hits.length(); i++) {
bw.write(
"<a href=\"file://"
+ hits.doc(i).get("descriptorImageIdentifier")
+ "\"><img src=\"file://"
+ hits.doc(i).get("descriptorImageIdentifier")
+ "\"></a><p>\n");
}
bw.write("</body>\n" + "</html>");
bw.close();
}
public void testSearchIndexLarge() throws IOException {
for (int i = 0; i < 10; i++) {
int queryDocID = (int) (Math.random() * 800);
// queryDocID = 877 * (i + 1);
IndexReader reader = DirectoryReader.open(FSDirectory.open(new File("index-large")));
// select one feature for the large index:
int featureIndex = 13;
int count = 0;
long ms = System.currentTimeMillis();
ImageSearchHits hits = searchers[featureIndex].search(reader.document(queryDocID), reader);
for (int j = 0; j < hits.length(); j++) {
String fileName = hits.doc(j).getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0];
System.out.println(hits.score(j) + ": \t" + fileName);
}
// FileUtils.saveImageResultsToHtml("GeneralTest_testSearchIndexLarge_", hits,
// reader.document(10).getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0]);
FileUtils.saveImageResultsToPng(
"GeneralTest_testSearchIndexLarge_" + i + "_",
hits,
reader.document(queryDocID).getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0]);
}
}
public void testCreateAndSearchSmallIndex() throws IOException {
for (int i = 0, buildersLength = builders.length; i < buildersLength; i++) {
DocumentBuilder b = builders[i];
// create an index with a specific builder:
IndexWriter iw = LuceneUtils.createIndexWriter(indexPath + "-small", true);
for (String identifier : testFiles) {
Document doc =
b.createDocument(new FileInputStream(testFilesPath + identifier), identifier);
doc.add(new StoredField("video_file", "surgery1.mp4"));
doc.add(new StoredField("timestamp", "25"));
iw.addDocument(doc);
}
iw.close();
ImageSearcher s = searchers[i];
IndexReader reader = DirectoryReader.open(FSDirectory.open(new File(indexPath + "-small")));
for (int k = 0; k < reader.maxDoc(); k++) {
Document query = reader.document(k);
ImageSearchHits hits = s.search(query, reader);
for (int y = 0; y < hits.length(); y++) {
Document result = hits.doc(y);
if (y == 0) {
// check if the first result is the query:
assertEquals(
result.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0].equals(
query.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0]),
true);
System.out.println(result.getValues("video_file")[0]);
} else {
// check if they are ordered by distance:
assertEquals(hits.score(y) < hits.score(y - 1), true);
}
}
}
}
}
public void testClassify() throws IOException {
boolean weightByRank = true;
String[] classes = {
"2012", "beach", "food", "london", "music", "nature", "people", "sky", "travel", "wedding"
};
int k = 50;
// CONFIG
String fieldName = DocumentBuilder.FIELD_NAME_COLORLAYOUT;
LireFeature feature = new ColorLayout();
String indexPath = "E:\\acmgc-cl-idx";
System.out.println(
"Tests for feature "
+ fieldName
+ " with k="
+ k
+ " - weighting by rank sum: "
+ weightByRank);
System.out.println("========================================");
HashMap<String, Integer> tag2count = new HashMap<String, Integer>(k);
HashMap<String, Double> tag2weight = new HashMap<String, Double>(k);
int c = 9; // used for just one class ...
// for (int c = 0; c < 10; c++) {
String classIdentifier = classes[c];
String listFiles = "D:\\DataSets\\Yahoo-GC\\test\\" + classIdentifier + ".txt";
// INIT
int[] confusion = new int[10];
Arrays.fill(confusion, 0);
HashMap<String, Integer> class2id = new HashMap<String, Integer>(10);
for (int i = 0; i < classes.length; i++) class2id.put(classes[i], i);
BufferedReader br = new BufferedReader(new FileReader(listFiles));
String line;
IndexReader ir = DirectoryReader.open(MMapDirectory.open(new File(indexPath)));
// in-memory linear search
// ImageSearcher bis = new GenericFastImageSearcher(k, feature.getClass(), fieldName,
// true, ir);
// hashing based searcher
BitSamplingImageSearcher bis =
new BitSamplingImageSearcher(k, fieldName, fieldName + "_hash", feature, 1000);
ImageSearchHits hits;
int count = 0, countCorrect = 0;
long ms = System.currentTimeMillis();
while ((line = br.readLine()) != null) {
try {
tag2count.clear();
tag2weight.clear();
hits = bis.search(ImageIO.read(new File(line)), ir);
// set tag weights and counts.
for (int l = 0; l < k; l++) {
String tag = getTag(hits.doc(l));
if (tag2count.get(tag) == null) tag2count.put(tag, 1);
else tag2count.put(tag, tag2count.get(tag) + 1);
if (weightByRank) {
if (tag2weight.get(tag) == null) tag2weight.put(tag, (double) l);
else tag2weight.put(tag, (double) l + tag2weight.get(tag));
} else {
if (tag2weight.get(tag) == null) tag2weight.put(tag, Double.valueOf(hits.score(l)));
else tag2weight.put(tag, (double) l + hits.score(l));
}
}
// find class:
int maxCount = 0, maxima = 0;
String classifiedAs = null;
for (Iterator<String> tagIterator = tag2count.keySet().iterator();
tagIterator.hasNext(); ) {
String tag = tagIterator.next();
if (tag2count.get(tag) > maxCount) {
maxCount = tag2count.get(tag);
maxima = 1;
classifiedAs = tag;
} else if (tag2count.get(tag) == maxCount) {
maxima++;
}
}
// if there are two or more classes with the same number of results, then we take a look at
// the weights.
// else the class is alread given in classifiedAs.
if (maxima > 1) {
double minWeight = Double.MAX_VALUE;
for (Iterator<String> tagIterator = tag2count.keySet().iterator();
tagIterator.hasNext(); ) {
String tag = tagIterator.next();
if (tag2weight.get(tag) < minWeight) {
minWeight = tag2weight.get(tag);
classifiedAs = tag;
}
}
}
// if (tag2.equals(tag3)) tag1 = tag2;
count++;
if (classifiedAs.equals(classIdentifier)) countCorrect++;
// confusion:
confusion[class2id.get(classifiedAs)]++;
// System.out.printf("%10s (%4.3f, %10d, %4d)\n", classifiedAs, ((double)
// countCorrect / (double) count), count, (System.currentTimeMillis() - ms) / count);
} catch (Exception e) {
System.err.println(e.getMessage());
}
}
// System.out.println("Results for class " + classIdentifier);
System.out.printf("Class\tAvg. Precision\tCount Test Images\tms per test\n");
System.out.printf(
"%s\t%4.5f\t%10d\t%4d\n",
classIdentifier,
((double) countCorrect / (double) count),
count,
(System.currentTimeMillis() - ms) / count);
System.out.printf("Confusion\t");
// for (int i = 0; i < classes.length; i++) {
// System.out.printf("%s\t", classes[i]);
// }
// System.out.println();
for (int i = 0; i < classes.length; i++) {
System.out.printf("%d\t", confusion[i]);
}
System.out.println();
// }
}
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();
}