Plotting

Precision vs recall curves for each evaluation run are generated by default and stored in the results subfolder of the model output directory.

To generate a comparison chart of various models use the compare command and pass as arguments the output paths of the models you would like to plot.

E.g.:

bob binseg compare -o myoutput -l myoutput/DRIVE/M2U-Net myoutput/DRIVE/U-Net myoutput/DRIVE/HED -t MyPlotTitle

Use bob binseg compare --help for more information.