baby.segmentation¶
Functions
|
Map optimisation-space radial spline params to standard values |
|
Weighted residual for radial spline optimisation |
|
Returns True if the regions have overlapping bounding boxes |
|
Get square-connected edges from filled image: |
|
|
|
|
|
Evaluate a radial spline defined by radii and angles rho: vector of radii for each point defining the spline phi: angles in [-pi,pi) defining points of the spline The spline is periodic across the boundary |
|
Format lists to return, depending on whether we want to return also masks and coordinates or not. |
|
|
|
|
|
|
|
Find regions in a probability image sorted by likelihood |
|
|
|
|
|
Max of intersection over a or over b |
|
Intersection over union (IoU) between boolean masks |
|
|
|
Segment cell outlines from morphology output of CNN as region ellipses |
|
|
|
Segment cell outlines from morphology output of CNN by fitting radial spline to threshold output |
|
Morphological segmentation for model predictions of flattener targets |
|
|
|
Segment cell outlines from morphology output of CNN by thresholding |
|
|
|
Weights radial residuals to bias towards the initial guess |
|
Helper function to convert row-column coords to radial coords |
|
Refine initial radial spline by optimising to predicted edge |
|
|
|
|
|
Segment cell outlines from morphology output of CNN by fitting radial spline to threshold output |
|
Returns a subset of masks excluding any overlaps with ref_masks |