Class BatchSimilarityGroundTruth
java.lang.Object
cz.fidentis.analyst.gui.app.tools.BatchSimilarityGroundTruth
A class for testing the efficiency of batch processing algorithms.
The goal of this tool is to create a "ground truth" measurements for other optimization techniques
- All pairs of faces are taken one by one from the collection.
- ICP registration is performed for each pair, but only in one direction (the second to the first)
- Hausdorff distance for each pair is computed in both direction (A-B and B-A). HD uses POINT_TO_POINT strategy and absolute distances.
ICP computation time: 02:27:32,170 HD computation time: 02:08:34,001 Total computation time: 05:29:55,321Stats for 100 faces WITHOUT CROP:
ICP computation time: 02:23:06,495 HD computation time: 01:59:55,520 Total computation time: 05:16:46,154
-
Constructor Summary
Constructors -
Method Summary
Modifier and TypeMethodDescriptionstatic void
Main methodprotected static void
registerAndMeasure
(int i, int j, HumanFace priFace, HumanFace trFace, double[][] distances)
-
Constructor Details
-
BatchSimilarityGroundTruth
public BatchSimilarityGroundTruth()
-
-
Method Details
-
main
Main method- Parameters:
args
- Input arguments- Throws:
IOException
- on IO errorClassNotFoundException
-
registerAndMeasure
-