Problem:
Currently, seed sorting is performed after seed detection, meaning detected seeds are stored unsorted in the SpatiotemporalAnalysis object. Every time object extraction is started (using existent seeds), the sorting is repeated. For real data, this can take several hours (depending on the number of seeds and scoring funtion).
For common applications / usage, we can assume the sorting is linked to the seed detection and needs no repetition for segmentation (only if the entire seed detection is repeated). I therefore suggest to perform the sorting before storing the seeds, i.e. updating the indices.
Solution:This step should be performed already here.
@dokempf do you see an issue or pitfall here? If not, I can implement the proposed adaptation and include in my testing.
Problem: Currently, seed sorting is performed after seed detection, meaning detected seeds are stored unsorted in the SpatiotemporalAnalysis object. Every time object extraction is started (using existent seeds), the sorting is repeated. For real data, this can take several hours (depending on the number of seeds and scoring funtion).
For common applications / usage, we can assume the sorting is linked to the seed detection and needs no repetition for segmentation (only if the entire seed detection is repeated). I therefore suggest to perform the sorting before storing the seeds, i.e. updating the indices.
Solution: This step should be performed already here.
@dokempf do you see an issue or pitfall here? If not, I can implement the proposed adaptation and include in my testing.