@Test
public void initialLearning() {
DummyVariance variance = new DummyVariance();
DummyFern fern = new DummyFern();
DummyTemplate template = new DummyTemplate();
DummyDetection detection = new DummyDetection();
TldLearning alg = new TldLearning(rand, config, template, variance, fern, detection);
FastQueue<ImageRectangle> regions = new FastQueue<ImageRectangle>(ImageRectangle.class, true);
regions.grow();
regions.grow();
regions.grow();
alg.initialLearning(new RectangleCorner2D_F64(10, 20, 30, 40), regions);
// Check to see if the variance threshold was set
assertEquals(1, variance.calledSelect);
// There should be a positive example
assertEquals(1, fern.calledP);
assertEquals(1, template.calledP);
// several negative examples too
assertEquals(3, fern.calledN);
// only negative template for ambiguous, which there are none since I'm being lazy
assertEquals(0, template.calledN);
assertEquals(1, detection.calledDetection);
}
/**
* Starts tracking the rectangular region.
*
* @param image First image in the sequence.
* @param x0 Top-left corner of rectangle. x-axis
* @param y0 Top-left corner of rectangle. y-axis
* @param x1 Bottom-right corner of rectangle. x-axis
* @param y1 Bottom-right corner of rectangle. y-axis
*/
public void initialize(T image, int x0, int y0, int x1, int y1) {
if (imagePyramid == null
|| imagePyramid.getInputWidth() != image.width
|| imagePyramid.getInputHeight() != image.height) {
int minSize = (config.trackerFeatureRadius * 2 + 1) * 5;
int scales[] = selectPyramidScale(image.width, image.height, minSize);
imagePyramid =
FactoryPyramid.discreteGaussian(scales, -1, 1, true, (Class<T>) image.getClass());
}
imagePyramid.process(image);
reacquiring = false;
targetRegion.set(x0, y0, x1, y1);
createCascadeRegion(image.width, image.height);
template.reset();
fern.reset();
tracking.initialize(imagePyramid);
variance.setImage(image);
template.setImage(image);
fern.setImage(image);
adjustRegion.init(image.width, image.height);
learning.initialLearning(targetRegion, cascadeRegions);
strongMatch = true;
previousTrackArea = targetRegion.area();
}
/**
* Updates track region.
*
* @param image Next image in the sequence.
* @return true if the object could be found and false if not
*/
public boolean track(T image) {
boolean success = true;
valid = false;
imagePyramid.process(image);
template.setImage(image);
variance.setImage(image);
fern.setImage(image);
if (reacquiring) {
// It can reinitialize if there is a single detection
detection.detectionCascade(cascadeRegions);
if (detection.isSuccess() && !detection.isAmbiguous()) {
TldRegion region = detection.getBest();
reacquiring = false;
valid = false;
// set it to the detected region
ImageRectangle r = region.rect;
targetRegion.set(r.x0, r.y0, r.x1, r.y1);
// get tracking running again
tracking.initialize(imagePyramid);
checkNewTrackStrong(region.confidence);
} else {
success = false;
}
} else {
detection.detectionCascade(cascadeRegions);
// update the previous track region using the tracker
trackerRegion.set(targetRegion);
boolean trackingWorked = tracking.process(imagePyramid, trackerRegion);
trackingWorked &= adjustRegion.process(tracking.getPairs(), trackerRegion);
TldHelperFunctions.convertRegion(trackerRegion, trackerRegion_I32);
if (hypothesisFusion(trackingWorked, detection.isSuccess())) {
// if it found a hypothesis and it is valid for learning, then learn
if (valid && performLearning) {
learning.updateLearning(targetRegion);
}
} else {
reacquiring = true;
success = false;
}
}
if (strongMatch) {
previousTrackArea = targetRegion.area();
}
return success;
}
@Test
public void updateLearning() {
DummyVariance variance = new DummyVariance();
DummyFern fern = new DummyFern();
DummyTemplate template = new DummyTemplate();
DummyDetection detection = new DummyDetection();
TldLearning alg = new TldLearning(rand, config, template, variance, fern, detection);
alg.updateLearning(new RectangleCorner2D_F64(10, 20, 30, 40));
// Check to see if the variance threshold was set
assertEquals(0, variance.calledSelect);
// There should be a positive example
assertEquals(1, fern.calledP);
assertEquals(1, template.calledP);
// several negative examples too
assertEquals(10, fern.calledN);
// only negative template for ambiguous, which there are none since I'm being lazy
assertEquals(0, template.calledN);
assertEquals(0, detection.calledDetection);
}
