search

trackmate-sc / TrackMate

TrackMate is your buddy for your everyday tracking.
https://imagej.net/plugins/trackmate
GNU General Public License v3.0
169 stars 75 forks source link

How about showing the detected blobs with the estimated diameter as radius? #5

Closed dscho closed 10 years ago

dscho commented 10 years ago

We should probably make this one optional and let the user choose the mode in the options dialog (which appears when double-clicking the tool icon) -- and maybe also let the user choose the color coding there?

dscho commented 10 years ago

Ooops. Duplicate of https://github.com/fiji/TrackMate/issues/4.

tinevez commented 10 years ago

I have found the estimate returned to be not robust enough. It performs well when spots are far from one another. But if not, if can return completely aberrant results. I therefore did not push this forward. However, if you want this happen there is an action that does that (in the latest pane of the GUI): just pick the action named 'Set radius to estimated value' to copy the estimated value to the actual radius feature.

A user reported me that the estimate was not robust enough for his application. I think that one robust way to do it would be to actually use the scale-space approach used in SIFT, at a computational cost of course.

dscho commented 10 years ago

I would like to have the option of assessing visually whether it is a good-enough measure to allow filtering based on the estimated diameter, therefore I would like to be able to toggle the displayed radius substantially earlier... (i.e. before I add that filter).

tinevez commented 10 years ago

It's not worth it. A good estimator might be worth the effort, but this one is not. This comes from user feedback and myself.

dscho commented 10 years ago

@tinevez then we should disable the "by estimated diameter" filter by now, no?

However, I'd rather provide a good estimator (or even better: an extensible mechanism to provide good estimators) because in the current scientific project I am involved in, there are wonderful, huge blobs. They are the things we do not want to track. And I have a hunch that the current estimator would be good enough (and the feature I'd like to implement would help...)

tinevez commented 10 years ago

Understood: if we have the scientific project, then we have to means to assess if it's good. What scope do you want for this issue? v2.2.0 (days, hopefully ;) or v2.3.0?

dscho commented 10 years ago

@tinevez do not let me hold up v2.2.0. If I make it until then, good. If not, please do not hesitate to go ahead and release 2.2.0.

tinevez commented 10 years ago

Assigned to v2.3.0 milestone.

tinevez commented 10 years ago

@dscho Let's close this an reopen it later?

dscho commented 10 years ago

Agreed.