Skip to content

Latest commit

 

History

History
28 lines (23 loc) · 1.94 KB

README.md

File metadata and controls

28 lines (23 loc) · 1.94 KB

Week 3. Post-processing techniques for background subtraction

Task 1. Hole filling to complete objects in the foreground.

  • Set the boolean 'task1' to True in the main.py and run.
  • Set 'holeFilling' to True and the rest to False in the postprocessing booleans.
  • backgroundEstimationAdaptive.optimalAlphaAdaptive() is executed for each dataset to find the AUC.

Task 2. Area filtering to remove noise from the background.

  • Set the boolean 'task2' to True in the main.py and run.
  • Set 'holeFilling' and 'areaFiltering' to True and the rest to False in the postprocessing booleans.
  • backgroundEstimationAdaptive.optimalPAdaptive() is executed to find the optimal P for each sequence.

Task 3. Morphological operators (closing + hole filling) to boost perfromance.

  • Set the boolean 'task3' to True in the main.py and run.
  • Set 'holeFilling', 'areaFiltering' and 'Morph' to True and the rest to False in the postprocessing booleans.
  • backgroundEstimationAdaptive.optimalAlphaAdaptive() is executed for each dataset to find the AUC.
  • All post-processing techniques up to task 3 are implemented in AdaptiveClassifier.postProcessing().

Task 4. Shadow detection and removal (pixel based methods using the HSV colorspace).

  • Set the boolean 'task4' to True in the main.py and run.
  • Set 'holeFilling', 'areaFiltering', 'Morph' and 'shadRemov' to True in the postprocessing booleans.
  • Shadow removal is implemented in AdaptiveClassifier.shadowRemoval().
  • To use method 1: uncomment lines 99-103 and comment lines 110-116.
  • To use method 2: comment lines 99-103 and uncomment lines 110-116.

Task 4. Improvement in Precision-Recall curves with respect to the best configuration from week 2.

  • Set the boolean 'task5' to True in the main.py and run.
  • readInputWriteOutput.precisionRecallCurveDataset() is used to plot the PR curves of each dataset.