locpix.scripts.img_seg.membrane_performance#

Module for evluating performance of membrane segmentation

For the train set, it will use the probability map over the image, and use this to assign probability of belonging to membrane to each localisation. Then it will use this to plot precision-recall curve, and calculate the optimal threshold.

For the test set, it will use the probability maps to plot a precision recall curve. It will then use the threshold calculated from the train set to threshold the test images

Now we have output membrane segmentations on the test set which we can use to finally calculate the other remaining performance metrics.

Functions

main()