Having a mask from semantic segmentation I want to split it to list of non-overlapping instances masks(like an output from instance segmentation). Do you know some fast algorithm, approach, tool or library to do this? I have found one python library. But it is very slow. Naive python implementation will also be slow.
$\begingroup$
$\endgroup$
2
-
$\begingroup$ Can you specify the library? Maybe edit your question by including a link pointing to it. $\endgroup$– Anoop A NairDec 14, 2020 at 14:43
-
$\begingroup$ Could you provide at least a sample image of the label to see if it's even feasible to do it? Instance Segmentation to Semantic Segmentation is possible, but this is not a reversible operation unless you can separate the instances by using some morphological operation or trained object detector. $\endgroup$– Pedro Henrique MonforteOct 30, 2021 at 5:00
Add a comment
|