Name Last Update
..
color_tables Loading commit data...
parameters_files Loading commit data...
alcd_labellisation_posttreatment.py Loading commit data...
all_run_pcc.py Loading commit data...
comparison.py Loading commit data...
find_chain_directory_paths.py Loading commit data...
masks_conversion.py Loading commit data...
metrics_grapher.py Loading commit data...
pixels_features_analysis.py Loading commit data...
png_converter.py Loading commit data...
readme.md Loading commit data...
statistics_synthesis.py Loading commit data...

readme.md

Processing Chains Comparator (PCC)

This tools allows the user to compare the output of cloud detection processing chains against a reference. The reference is created with ALCD beforehand. The current supported processing chains (PC) are MAJA, Sen2Cor and Fmask

This readme describes quickly the use of each python file, the general use of the tool is described in the user manual in the parent directory.

Description of each file

  • all_run_pcc.py main file, calls the other files. Run the full PCC code.
    Called by: /
  • masks_conversion.py used to convert the output from the 3 PC to the ALCD equivalent one.
    Called by: all_run_pcc.py
  • find_chain_directory_paths.py to search for the paths of the chains outputs.
    Called by: all_run_pcc.py
  • comparison.py compares the converted masks output by the PCs against the ALCD reference.
    Called by: all_run_pcc.py
  • metrics_grapher.py plots the metrics (accuracy, f1-score, recall, precision) for a scene, with the condition that the 3 PCs have been run on it.
    Called by: all_run_pcc.py
  • alcd_labellisation_posttreatment.py used to perform morphological operations on labeled GTiffs
    Called by: masks_conversion.py
  • png_converter.py converts Gtiffs to pngs, and adds a text to them (PC name, location and date of the scene)
    Called by: all_run_pcc.py
  • statistics_synthesis.py gives global statistics on all the scenes.
    Called by: /
  • pixels_features_analysis.py analyses the features values of the original image, given their ALCD reference class. It is now deprecated.
    Called by: all_run_pcc.py

Getting Started

You need to set the parameters as described in the user manual. Then you can simply run

python all_run_pcc.py -l Arles -d 20171002 -b True

Once you did it for all the scenes, you can aggregate the results with

python statistics_synthesis.py

Outputs

For each scene, a root_scene/ output dir will be created. In this directory you will find after having run PCC, among other things:

  • root_scene/Binary_classif/ --- contains the conversion of the original masks to the binary valid/invalid classification
  • root_scene/Binary_difference/ --- contains the comparison of the binary masks against the ALCD reference
  • root_scene/Multi_classif/ --- contains the conversion of the original masks to the ALCD multi-class classification equivalent
  • root_scene/Multi_difference/ --- contains the comparison of the multi-class masks against the ALCD reference
  • root_scene/Original_data/ --- contains the original masks, realigned and resampled to the desired resolution
  • root_scene/Out/ --- contains the .png of the binary differences, the quicklook from the original image and the contours of the ALCD reference overlayed on it
  • root_scene/Statistics/ --- contains the statistics from the differences