Example Scene Classification - Revision history

Peter at 23:14, 17 January 2022

2022-01-17T23:14:43Z

← Older revision		Revision as of 23:14, 17 January 2022
Line 5:		Line 5:
	</center>		</center>

	Scene classification is the problem where you are presented with an image and you need to classify it as belonging to a known set. The example below demonstrates how to perform scene classification using dense SURF features and a nearest neighbour classifier. See code documentation for a more detailed discussion.		Scene classification is the problem where you are presented with an image and you need to classify it as belonging to a known set. The example below demonstrates how to perform scene classification using dense SURF features and a nearest neighbour classifier. See code documentation for a more detailed discussion.

	Example Code:		Example Code:
	* [https://github.com/lessthanoptimal/BoofCV/blob/v0.38/examples/src/main/java/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]		* [https://github.com/lessthanoptimal/BoofCV/blob/v0.40/examples/src/main/java/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]

	Concepts:		Concepts:
Line 94:		Line 94:

	// Use these clusters to assign features to words		// Use these clusters to assign features to words
	~~FeatureToWordHistogram_F64~~ featuresToHistogram = new FeatureToWordHistogram_F64(assignment, HISTOGRAM_HARD);		var featuresToHistogram = new FeatureToWordHistogram_F64(assignment, HISTOGRAM_HARD);

	// Storage for the work histogram in each image in the training set and their label		// Storage for the work histogram in each image in the training set and their label
Line 112:		Line 112:
	System.out.println("Image Features");		System.out.println("Image Features");

	// ~~computes~~ features in the training image set		// Compute features in the training image set
	~~List<TupleDesc_F64>~~ features = new ArrayList<>();		var features = new ArrayList<TupleDesc_F64>();
	for (String scene : train.keySet()) {		for (String scene : train.keySet()) {
	List<String> imagePaths = train.get(scene);		List<String> imagePaths = train.get(scene);
Line 147:		Line 147:
	AssignCluster<double[]> assignment = UtilIO.load(CLUSTER_FILE_NAME);		AssignCluster<double[]> assignment = UtilIO.load(CLUSTER_FILE_NAME);

	~~FeatureToWordHistogram_F64~~ featuresToHistogram = new FeatureToWordHistogram_F64(assignment, HISTOGRAM_HARD);		var featuresToHistogram = new FeatureToWordHistogram_F64(assignment, HISTOGRAM_HARD);


Line 209:		Line 209:
	public static void main( String[] args ) {		public static void main( String[] args ) {

	~~ConfigDenseSurfFast~~ surfFast = new ConfigDenseSurfFast(new DenseSampling(8, 8));		var surfFast = new ConfigDenseSurfFast(new DenseSampling(8, 8));
	// ConfigDenseSurfStable surfStable = new ConfigDenseSurfStable(new DenseSampling(8,8));		// ConfigDenseSurfStable surfStable = new ConfigDenseSurfStable(new DenseSampling(8,8));
	// ConfigDenseSift sift = new ConfigDenseSift(new DenseSampling(6,6));		// ConfigDenseSift sift = new ConfigDenseSift(new DenseSampling(6,6));
Line 220:		Line 220:
	// FactoryDescribeImageDense.hog(hog, ImageType.single(GrayU8.class));		// FactoryDescribeImageDense.hog(hog, ImageType.single(GrayU8.class));

	~~ConfigKMeans~~ configKMeans = new ConfigKMeans();		var configKMeans = new ConfigKMeans();
	configKMeans.maxIterations = MAX_KNN_ITERATIONS;		configKMeans.maxIterations = MAX_KNN_ITERATIONS;
	configKMeans.reseedAfterIterations = 20;		configKMeans.reseedAfterIterations = 20;
	ComputeClusters<double[]> clusterer = FactoryClustering.~~kMeans~~(		ComputeClusters<double[]> clusterer = FactoryClustering.kMeans_MT(
	configKMeans, desc.createDescription().size(), double[].class);		configKMeans, desc.createDescription().size(), 200, double[].class);
	clusterer.setVerbose(true);		clusterer.setVerbose(true);
			// The _MT tells it to use the threaded version. This can run MUCH faster.

	int pointDof = desc.createDescription().size();		int pointDof = desc.createDescription().size();
Line 231:		Line 232:
	ExampleClassifySceneKnn example = new ExampleClassifySceneKnn(desc, clusterer, nn);		ExampleClassifySceneKnn example = new ExampleClassifySceneKnn(desc, clusterer, nn);

	~~File~~ trainingDir = new File(UtilIO.pathExample("learning/scene/train"));		var trainingDir = new File(UtilIO.pathExample("learning/scene/train"));
	~~File~~ testingDir = new File(UtilIO.pathExample("learning/scene/test"));		var testingDir = new File(UtilIO.pathExample("learning/scene/test"));

	if (!trainingDir.exists() \|\| !testingDir.exists()) {		if (!trainingDir.exists() \|\| !testingDir.exists()) {
Line 279:		Line 280:
	}		}
	}		}

	</syntaxhighlight>		</syntaxhighlight>

Peter at 19:07, 12 July 2021

2021-07-12T19:07:48Z

← Older revision		Revision as of 19:07, 12 July 2021
Line 8:		Line 8:

	Example Code:		Example Code:
	* [https://github.com/lessthanoptimal/BoofCV/blob/v0.37/examples/src/main/java/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]		* [https://github.com/lessthanoptimal/BoofCV/blob/v0.38/examples/src/main/java/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]

	Concepts:		Concepts:
Line 24:		Line 24:
	/**		/**
	* <p>		* <p>
	* Example of how to train a K-NN bow-of-word classifier for scene recognition. The resulting classifier		* Example of how to train a K-NN bow-of-word classifier for scene recognition. The resulting classifier
	* produces results which are correct 52.2% of the time. To provide a point of comparison, randomly selecting		* produces results which are correct 52.2% of the time. To provide a point of comparison, randomly selecting
	* a scene is about 6.7% accurate, SVM One vs One RBF classifier can produce accuracy of around 74% and		* a scene is about 6.7% accurate, SVM One vs One RBF classifier can produce accuracy of around 74% and
	* other people using different techniques claim to have achieved around 85% accurate with more advanced		* other people using different techniques claim to have achieved around 85% accurate with more advanced
Line 74:		Line 74:
	NearestNeighbor<HistogramScene> nn ) {		NearestNeighbor<HistogramScene> nn ) {
	this.describeImage = describeImage;		this.describeImage = describeImage;
	this.cluster = new ClusterVisualWords(clusterer~~, describeImage.createDescription().size()~~, 0xFEEDBEEF);		this.cluster = new ClusterVisualWords(clusterer, 0xFEEDBEEF);
	this.nn = nn;		this.nn = nn;
	}		}

	/**		/**
	* Process all the data in the training data set to learn the classifications. See code for details.		* Process all the data in the training data set to learn the classifications. See code for details.
	*/		*/
	public void learnAndSave() {		public void learnAndSave() {
Line 107:		Line 107:

	/**		/**
	* Extract dense features across the training set. Then clusters are found within those features.		* Extract dense features across the training set. Then clusters are found within those features.
	*/		*/
	private AssignCluster<double[]> computeClusters() {		private AssignCluster<double[]> computeClusters() {
Line 134:		Line 134:

	System.out.println("Clustering");		System.out.println("Clustering");
	// Find the clusters. This can take a bit		// Find the clusters. This can take a bit
	cluster.process(NUMBER_OF_WORDS);		cluster.process(NUMBER_OF_WORDS);

Line 159:		Line 159:

	/**		/**
	* For all the images in the training data set it computes a {@link HistogramScene}. That data structure		* For all the images in the training data set it computes a {@link HistogramScene}. That data structure
	* contains the word histogram and the scene that the histogram belongs to.		* contains the word histogram and the scene that the histogram belongs to.
	*/		*/
Line 185:		Line 185:
	featuresToHistogram.process();		featuresToHistogram.process();

	// The histogram is already normalized so that it sums up to 1. This provides invariance		// The histogram is already normalized so that it sums up to 1. This provides invariance
	// against the overall number of features changing.		// against the overall number of features changing.
	double[] histogram = featuresToHistogram.getHistogram();		double[] histogram = featuresToHistogram.getHistogram();
Line 220:		Line 220:
	// FactoryDescribeImageDense.hog(hog, ImageType.single(GrayU8.class));		// FactoryDescribeImageDense.hog(hog, ImageType.single(GrayU8.class));

	ComputeClusters<double[]> clusterer = FactoryClustering.~~kMeans_F64~~(~~null~~, ~~MAX_KNN_ITERATIONS~~, ~~20, 1e-6~~);		ConfigKMeans configKMeans = new ConfigKMeans();
			configKMeans.maxIterations = MAX_KNN_ITERATIONS;
			configKMeans.reseedAfterIterations = 20;
			ComputeClusters<double[]> clusterer = FactoryClustering.kMeans(
			configKMeans, desc.createDescription().size(), double[].class);
	clusterer.setVerbose(true);		clusterer.setVerbose(true);

Line 258:		Line 262:

	// Show confusion matrix		// Show confusion matrix
	// Not the best coloration scheme... perfect = red diagonal and blue elsewhere.		// Not the best coloration scheme... perfect = red diagonal and blue elsewhere.
	ShowImages.showWindow(new ConfusionMatrixPanel(		ShowImages.showWindow(new ConfusionMatrixPanel(
	confusion.getMatrix(), example.getScenes(), 400, true), "Confusion Matrix", true);		confusion.getMatrix(), example.getScenes(), 400, true), "Confusion Matrix", true);
Line 269:		Line 273:
	// SURF results are interesting. "Stable" is significantly better than "fast"!		// SURF results are interesting. "Stable" is significantly better than "fast"!
	// One explanation is that the descriptor for "fast" samples a smaller region than "stable", by a		// One explanation is that the descriptor for "fast" samples a smaller region than "stable", by a
	// couple of pixels at scale of 1. Thus there is less overlap between the features.		// couple of pixels at scale of 1. Thus there is less overlap between the features.

	// Reducing the size of "stable" to 0.95 does slightly improve performance to 50.5%, can't scale it down		// Reducing the size of "stable" to 0.95 does slightly improve performance to 50.5%, can't scale it down

Peter at 02:32, 22 December 2020

2020-12-22T02:32:47Z

← Older revision		Revision as of 02:32, 22 December 2020
Line 8:		Line 8:

	Example Code:		Example Code:
	* [https://github.com/lessthanoptimal/BoofCV/blob/v0.31/examples/src/main/java/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]		* [https://github.com/lessthanoptimal/BoofCV/blob/v0.37/examples/src/main/java/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]

	Concepts:		Concepts:
Line 65:		Line 65:
	// Algorithms		// Algorithms
	ClusterVisualWords cluster;		ClusterVisualWords cluster;
	DescribeImageDense<GrayU8,TupleDesc_F64> describeImage;		DescribeImageDense<GrayU8, TupleDesc_F64> describeImage;
	NearestNeighbor<HistogramScene> nn;		NearestNeighbor<HistogramScene> nn;

	ClassifierKNearestNeighborsBow<GrayU8,TupleDesc_F64> classifier;		ClassifierKNearestNeighborsBow<GrayU8, TupleDesc_F64> classifier;

	public ExampleClassifySceneKnn(final DescribeImageDense<GrayU8, TupleDesc_F64> describeImage,		public ExampleClassifySceneKnn( final DescribeImageDense<GrayU8, TupleDesc_F64> describeImage,
	ComputeClusters<double[]> clusterer,		ComputeClusters<double[]> clusterer,
	NearestNeighbor<HistogramScene> nn) {		NearestNeighbor<HistogramScene> nn ) {
	this.describeImage = describeImage;		this.describeImage = describeImage;
	this.cluster = new ClusterVisualWords(clusterer, describeImage.createDescription().size(),0xFEEDBEEF);		this.cluster = new ClusterVisualWords(clusterer, describeImage.createDescription().size(), 0xFEEDBEEF);
	this.nn = nn;		this.nn = nn;
	}		}
Line 86:		Line 86:
	// Either load pre-computed words or compute the words from the training images		// Either load pre-computed words or compute the words from the training images
	AssignCluster<double[]> assignment;		AssignCluster<double[]> assignment;
	if( new File(CLUSTER_FILE_NAME).exists() ) {		if (new File(CLUSTER_FILE_NAME).exists()) {
	assignment = UtilIO.load(CLUSTER_FILE_NAME);		assignment = UtilIO.load(CLUSTER_FILE_NAME);
	} else {		} else {
Line 94:		Line 94:

	// Use these clusters to assign features to words		// Use these clusters to assign features to words
	FeatureToWordHistogram_F64 featuresToHistogram = new FeatureToWordHistogram_F64(assignment,HISTOGRAM_HARD);		FeatureToWordHistogram_F64 featuresToHistogram = new FeatureToWordHistogram_F64(assignment, HISTOGRAM_HARD);

	// Storage for the work histogram in each image in the training set and their label		// Storage for the work histogram in each image in the training set and their label
	List<HistogramScene> memory;		List<HistogramScene> memory;

	if( !new File(HISTOGRAM_FILE_NAME).exists() ) {		if (!new File(HISTOGRAM_FILE_NAME).exists()) {
	System.out.println(" computing histograms");		System.out.println(" computing histograms");
	memory = computeHistograms(featuresToHistogram);		memory = computeHistograms(featuresToHistogram);
	UtilIO.save(memory,HISTOGRAM_FILE_NAME);		UtilIO.save(memory, HISTOGRAM_FILE_NAME);
	}		}
	}		}
Line 114:		Line 114:
	// computes features in the training image set		// computes features in the training image set
	List<TupleDesc_F64> features = new ArrayList<>();		List<TupleDesc_F64> features = new ArrayList<>();
	for( String scene : train.keySet() ) {		for (String scene : train.keySet()) {
	List<String> imagePaths = train.get(scene);		List<String> imagePaths = train.get(scene);
	System.out.println(" " + scene);		System.out.println(" " + scene);

	for( String path : imagePaths ) {		for (String path : imagePaths) {
	GrayU8 image = UtilImageIO.loadImage(path, GrayU8.class);		GrayU8 image = UtilImageIO.loadImage(path, GrayU8.class);
	describeImage.process(image);		describeImage.process(image);

	// the descriptions will get recycled on the next call, so create a copy		// the descriptions will get recycled on the next call, so create a copy
	for( TupleDesc_F64 d : describeImage.getDescriptions() ) {		for (TupleDesc_F64 d : describeImage.getDescriptions()) {
	features.add( d.copy() );		features.add(d.copy());
	}		}
	}		}
Line 147:		Line 147:
	AssignCluster<double[]> assignment = UtilIO.load(CLUSTER_FILE_NAME);		AssignCluster<double[]> assignment = UtilIO.load(CLUSTER_FILE_NAME);

	FeatureToWordHistogram_F64 featuresToHistogram = new FeatureToWordHistogram_F64(assignment,HISTOGRAM_HARD);		FeatureToWordHistogram_F64 featuresToHistogram = new FeatureToWordHistogram_F64(assignment, HISTOGRAM_HARD);


Line 162:		Line 162:
	* contains the word histogram and the scene that the histogram belongs to.		* contains the word histogram and the scene that the histogram belongs to.
	*/		*/
	private List<HistogramScene> computeHistograms(FeatureToWordHistogram_F64 featuresToHistogram ) {		private List<HistogramScene> computeHistograms( FeatureToWordHistogram_F64 featuresToHistogram ) {

	List<String> scenes = getScenes();		List<String> scenes = getScenes();
Line 169:		Line 169:
	memory = new ArrayList<>();		memory = new ArrayList<>();

	for( int sceneIndex = 0; sceneIndex < scenes.size(); sceneIndex++ ) {		for (int sceneIndex = 0; sceneIndex < scenes.size(); sceneIndex++) {
	String scene = scenes.get(sceneIndex);		String scene = scenes.get(sceneIndex);
	System.out.println(" " + scene);		System.out.println(" " + scene);
Line 180:		Line 180:
	featuresToHistogram.reset();		featuresToHistogram.reset();
	describeImage.process(image);		describeImage.process(image);
	for ( TupleDesc_F64 d : describeImage.getDescriptions() ) {		for (TupleDesc_F64 d : describeImage.getDescriptions()) {
	featuresToHistogram.addFeature(d);		featuresToHistogram.addFeature(d);
	}		}
Line 201:		Line 201:

	@Override		@Override
	protected int classify(String path) {		protected int classify( String path ) {
	GrayU8 image = UtilImageIO.loadImage(path, GrayU8.class);		GrayU8 image = UtilImageIO.loadImage(path, GrayU8.class);

Line 207:		Line 207:
	}		}

	public static void main(String[] args) {		public static void main( String[] args ) {

	ConfigDenseSurfFast surfFast = new ConfigDenseSurfFast(new DenseSampling(8,8));		ConfigDenseSurfFast surfFast = new ConfigDenseSurfFast(new DenseSampling(8, 8));
	ConfigDenseSurfStable surfStable = new ConfigDenseSurfStable(new DenseSampling(8,8));		// ConfigDenseSurfStable surfStable = new ConfigDenseSurfStable(new DenseSampling(8,8));
	ConfigDenseSift sift = new ConfigDenseSift(new DenseSampling(6,6));		// ConfigDenseSift sift = new ConfigDenseSift(new DenseSampling(6,6));
	ConfigDenseHoG hog = new ConfigDenseHoG();		// ConfigDenseHoG hog = new ConfigDenseHoG();

	DescribeImageDense<GrayU8,TupleDesc_F64> desc = ~~(DescribeImageDense)~~		DescribeImageDense<GrayU8, TupleDesc_F64> desc =
	FactoryDescribeImageDense.surfFast(surfFast, GrayU8.class);		FactoryDescribeImageDense.surfFast(surfFast, GrayU8.class);
	// FactoryDescribeImageDense.surfStable(surfStable, GrayU8.class);		// FactoryDescribeImageDense.surfStable(surfStable, GrayU8.class);
Line 225:		Line 225:
	int pointDof = desc.createDescription().size();		int pointDof = desc.createDescription().size();
	NearestNeighbor<HistogramScene> nn = FactoryNearestNeighbor.exhaustive(new KdTreeHistogramScene_F64(pointDof));		NearestNeighbor<HistogramScene> nn = FactoryNearestNeighbor.exhaustive(new KdTreeHistogramScene_F64(pointDof));
	ExampleClassifySceneKnn example = new ExampleClassifySceneKnn(desc,clusterer,nn);		ExampleClassifySceneKnn example = new ExampleClassifySceneKnn(desc, clusterer, nn);

	File trainingDir = new File(UtilIO.pathExample("learning/scene/train"));		File trainingDir = new File(UtilIO.pathExample("learning/scene/train"));
	File testingDir = new File(UtilIO.pathExample("learning/scene/test"));		File testingDir = new File(UtilIO.pathExample("learning/scene/test"));

	if( !trainingDir.exists() \|\| !testingDir.exists() ) {		if (!trainingDir.exists() \|\| !testingDir.exists()) {
	String addressSrc = "http://boofcv.org/notwiki/largefiles/bow_data_v001.zip";		String addressSrc = "http://boofcv.org/notwiki/largefiles/bow_data_v001.zip";
	File dst = new File(trainingDir.getParentFile(),"bow_data_v001.zip");		File dst = new File(trainingDir.getParentFile(), "bow_data_v001.zip");
	try {		try {
	DeepBoofDataBaseOps.download(addressSrc, dst);		DeepBoofDataBaseOps.download(addressSrc, dst);
	DeepBoofDataBaseOps.decompressZip(dst, dst.getParentFile(), true);		DeepBoofDataBaseOps.decompressZip(dst, dst.getParentFile(), true);
	System.out.println("Download complete!");		System.out.println("Download complete!");
	} catch( IOException e ) {		} catch (IOException e) {
	throw new ~~RuntimeException~~(e);		throw new UncheckedIOException(e);
	}		}
	} else {		} else {
	System.out.println("Delete and download again if there are file not found errors");		System.out.println("Delete and download again if there are file not found errors");
	System.out.println(" "+trainingDir);		System.out.println(" " + trainingDir);
	System.out.println(" "+testingDir);		System.out.println(" " + testingDir);
	}		}

Line 260:		Line 260:
	// Not the best coloration scheme... perfect = red diagonal and blue elsewhere.		// Not the best coloration scheme... perfect = red diagonal and blue elsewhere.
	ShowImages.showWindow(new ConfusionMatrixPanel(		ShowImages.showWindow(new ConfusionMatrixPanel(
	confusion.getMatrix(),example.getScenes(), 400, true), "Confusion Matrix", true);		confusion.getMatrix(), example.getScenes(), 400, true), "Confusion Matrix", true);

	// For SIFT descriptor the accuracy is 54.0%		// For SIFT descriptor the accuracy is 54.0%

Peter at 02:58, 17 October 2018

2018-10-17T02:58:48Z

← Older revision		Revision as of 02:58, 17 October 2018
Line 8:		Line 8:

	Example Code:		Example Code:
	* [https://github.com/lessthanoptimal/BoofCV/blob/v0.27/examples/src/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]		* [https://github.com/lessthanoptimal/BoofCV/blob/v0.31/examples/src/main/java/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]

	Concepts:		Concepts:
Line 223:		Line 223:
	clusterer.setVerbose(true);		clusterer.setVerbose(true);

	NearestNeighbor<HistogramScene> nn = FactoryNearestNeighbor.exhaustive();		int pointDof = desc.createDescription().size();
			NearestNeighbor<HistogramScene> nn = FactoryNearestNeighbor.exhaustive(new KdTreeHistogramScene_F64(pointDof));
	ExampleClassifySceneKnn example = new ExampleClassifySceneKnn(desc,clusterer,nn);		ExampleClassifySceneKnn example = new ExampleClassifySceneKnn(desc,clusterer,nn);

Peter at 16:04, 17 August 2017

2017-08-17T16:04:25Z

← Older revision		Revision as of 16:04, 17 August 2017
Line 8:		Line 8:

	Example Code:		Example Code:
	* [https://github.com/lessthanoptimal/BoofCV/blob/v0.25/examples/src/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]		* [https://github.com/lessthanoptimal/BoofCV/blob/v0.27/examples/src/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]

	Concepts:		Concepts:
Line 230:		Line 230:

	if( !trainingDir.exists() \|\| !testingDir.exists() ) {		if( !trainingDir.exists() \|\| !testingDir.exists() ) {
	String addressSrc = "http://~~jaist.dl.sourceforge~~.~~net/project~~/~~boofcv~~/~~datasets~~/bow_data_v001.zip";		String addressSrc = "http://boofcv.org/notwiki/largefiles/bow_data_v001.zip";
	File dst = trainingDir.getParentFile();		File dst = new File(trainingDir.getParentFile(),"bow_data_v001.zip");
	DeepBoofDataBaseOps.download(addressSrc,dst);		try {
	DeepBoofDataBaseOps.decompressZip~~(new File~~(dst, ~~"bow_data_v001~~.~~zip"~~)~~, dst~~, true);		DeepBoofDataBaseOps.download(addressSrc, dst);
	System.out.println("Download complete!");		DeepBoofDataBaseOps.decompressZip(dst, dst.getParentFile(), true);
			System.out.println("Download complete!");
			} catch( IOException e ) {
			throw new RuntimeException(e);
			}
	} else {		} else {
	System.out.println("Delete and download again if there are file not found errors");		System.out.println("Delete and download again if there are file not found errors");

Peter at 05:30, 2 December 2016

2016-12-02T05:30:50Z

← Older revision		Revision as of 05:30, 2 December 2016
Line 15:		Line 15:
	* Clustering		* Clustering
	* k-NN classifier		* k-NN classifier

			Related Examples:
			* [[Example_Image_Classification\|Image Classification]]

	= Example Code =		= Example Code =

Peter at 00:09, 2 December 2016

2016-12-02T00:09:53Z

← Older revision		Revision as of 00:09, 2 December 2016
Line 8:		Line 8:

	Example Code:		Example Code:
	* [https://github.com/lessthanoptimal/BoofCV/blob/v0.24/examples/src/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]		* [https://github.com/lessthanoptimal/BoofCV/blob/v0.25/examples/src/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]

	Concepts:		Concepts:
Line 110:		Line 110:

	// computes features in the training image set		// computes features in the training image set
	List<TupleDesc_F64> features = new ArrayList<~~TupleDesc_F64~~>();		List<TupleDesc_F64> features = new ArrayList<>();
	for( String scene : train.keySet() ) {		for( String scene : train.keySet() ) {
	List<String> imagePaths = train.get(scene);		List<String> imagePaths = train.get(scene);
Line 150:		Line 150:
	// Can use this classifier with saved results and avoid the		// Can use this classifier with saved results and avoid the

	classifier = new ClassifierKNearestNeighborsBow<~~GrayU8,TupleDesc_F64~~>(nn,describeImage,featuresToHistogram);		classifier = new ClassifierKNearestNeighborsBow<>(nn, describeImage, featuresToHistogram);
	classifier.setClassificationData(memory, getScenes().size());		classifier.setClassificationData(memory, getScenes().size());
	classifier.setNumNeighbors(NUM_NEIGHBORS);		classifier.setNumNeighbors(NUM_NEIGHBORS);
Line 164:		Line 164:

	List<HistogramScene> memory;// Processed results which will be passed into the k-NN algorithm		List<HistogramScene> memory;// Processed results which will be passed into the k-NN algorithm
	memory = new ArrayList<~~HistogramScene~~>();		memory = new ArrayList<>();

	for( int sceneIndex = 0; sceneIndex < scenes.size(); sceneIndex++ ) {		for( int sceneIndex = 0; sceneIndex < scenes.size(); sceneIndex++ ) {
Line 227:		Line 227:

	if( !trainingDir.exists() \|\| !testingDir.exists() ) {		if( !trainingDir.exists() \|\| !testingDir.exists() ) {
	String ~~path~~ = ~~UtilIO~~.~~pathExample~~("~~learning/scene/~~");		String addressSrc = "http://jaist.dl.sourceforge.net/project/boofcv/datasets/bow_data_v001.zip";
	System.~~err~~.println("~~Please follow instructions in~~ "~~+path+~~" and download ~~the~~");		File dst = trainingDir.getParentFile();
	System.~~err~~.println("~~required files~~");		DeepBoofDataBaseOps.download(addressSrc,dst);
	System.~~exit~~(1);		DeepBoofDataBaseOps.decompressZip(new File(dst, "bow_data_v001.zip"), dst, true);
			System.out.println("Download complete!");
			} else {
			System.out.println("Delete and download again if there are file not found errors");
			System.out.println(" "+trainingDir);
			System.out.println(" "+testingDir);
	}		}

Line 246:		Line 251:
	// Show confusion matrix		// Show confusion matrix
	// Not the best coloration scheme... perfect = red diagonal and blue elsewhere.		// Not the best coloration scheme... perfect = red diagonal and blue elsewhere.
	ShowImages.showWindow(new ConfusionMatrixPanel(confusion.getMatrix(), 400, true), "Confusion Matrix", true);		ShowImages.showWindow(new ConfusionMatrixPanel(
			confusion.getMatrix(),example.getScenes(), 400, true), "Confusion Matrix", true);

	// For SIFT descriptor the accuracy is 54.0%		// For SIFT descriptor the accuracy is 54.0%

Peter at 19:04, 4 June 2016

2016-06-04T19:04:52Z

← Older revision		Revision as of 19:04, 4 June 2016
Line 1:		Line 1:
	<center>		<center>
	<gallery widths=~~300px~~ heights=~~300px~~>		<gallery widths=450px heights=340px>
	File:Confusion_matrix.png \| Confusion matrix created from the code below. Perfect results would be black along the diagonal.		File:Confusion_matrix.png \| Confusion matrix created from the code below. Perfect results would be black along the diagonal.
	</gallery>		</gallery>

Peter at 19:04, 4 June 2016

2016-06-04T19:04:11Z

← Older revision		Revision as of 19:04, 4 June 2016
Line 1:		Line 1:
	<center>		<center>
	<gallery widths=~~450px~~ heights=~~325px~~>		<gallery widths=300px heights=300px>
	File:Confusion_matrix.png \| Confusion matrix created from the code below. Perfect results would be black along the diagonal.		File:Confusion_matrix.png \| Confusion matrix created from the code below. Perfect results would be black along the diagonal.
	</gallery>		</gallery>
Line 8:		Line 8:

	Example Code:		Example Code:
	* [https://github.com/lessthanoptimal/BoofCV/blob/v0.23/examples/src/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]		* [https://github.com/lessthanoptimal/BoofCV/blob/v0.24/examples/src/boofcv/examples/recognition/ExampleClassifySceneKnn.java ExampleClassifySceneKnn.java ]

	Concepts:		Concepts:

Peter at 18:58, 4 June 2016

2016-06-04T18:58:54Z

← Older revision		Revision as of 18:58, 4 June 2016
Line 1:		Line 1:
	<center>		<center>
	<gallery widths=~~300px~~ heights=~~300px~~>		<gallery widths=450px heights=325px>
	File:Confusion_matrix.png \| Confusion matrix created from the code below. Perfect results would be black along the diagonal.		File:Confusion_matrix.png \| Confusion matrix created from the code below. Perfect results would be black along the diagonal.
	</gallery>		</gallery>