-
Notifications
You must be signed in to change notification settings - Fork 34
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
CFU per ROI #1
Comments
Howdy: Was this ever implemented? I have my petri plates set up into six sections that I'd like counted (I can easily get >100 colonies a section). As of now, it seems likes any colony which falls under a ROI receives a "ROI=1" tag but doesn't get differentiated between ROIs. Cheers |
I am pretty sure I implemented this, but it sounds like either I didn't, or there is a bug. I'll try and reproduce the error tonight and get back to you with a fix. |
It appears to work with 2 ROIs but not more than two. I'll try and fool Clarity Edit: When I have six ROIs, their results are grouped into two ROIs in the summary page. So 3 of my ROIs are "ROI 1" and the other 3 ROIs are "ROI 2" On Mon, Nov 21, 2016 at 7:43 AM, Nathanael Lampe notifications@github.com
|
If you can send the image you're trying to classify, and a screenshot of the ROI's you're setting up I can try and reproduce the error. It's worth noting that the clustering (as far as I remember) does not operate on individual ROI's, but on the ensemble of colonies that are valid (that is, it uses red colonies across all ROIs and yellow colonies across all ROIs to determine what is the extent of the red and yellow clusters, for example). Just checking also that the ROI numbers return to normal when you disable the colour recognition? |
You can also try to replicate the bug on the sample image two-colours.jpg This is what I get with three ROIs, unless I'm missing something I can't reproduce your bug: I tried the same image with 7 ROI's and the summary page in the GUI recorded 7 ROI's, and saved the summary information correctly (unless I'm missing something). A dump of the summary data for 7 ROI's is below, let me know if I'm missing something obvious.
|
I'm on a WiFi Hotspot as I commute into work. I'll upload my image when I On Nov 21, 2016 8:23 AM, "Nathanael Lampe" notifications@github.com wrote:
|
That seems pretty bad. Thanks for reporting the issue @danieljnilson. I think we should try to see how independent from the GUI front-end the problem is. For instance, use a mask file and the CLI to see if we can reproduce this. |
@danieljnilson have you compiled OpenCFU from the git (master), or you are using your OS software repo? |
I am using the compiled git (master). I tried downloading from my OS software repo (Kubuntu) but it couldn't be found... Later on today if I have time, I'll make a ROI mask and try to use it through the CLI. If not, I'll experiment it with at around 8-9 PM PST. Cheers, I appreciate you looking into this so readily. Edit: Kubuntu 16.10 |
I tried to reproduce the error but I couldn't, neither with circle selections nor with convex hulls inside the GUI: Is there something I'm missing in how I'm trying to reproduce the bug? This is probably one for @qgeissmann to look into rather than me, but if it ends up being because of the colour clustering code somehow, I'll do what I can. I am compiling OpenCFU against OpenCV 2, in case that helps. |
No, you're (seemingly) doing exactly what I am doing. Now that I've been working on my raw data (over 100 image files, woo), I notice I can't consistently recreate the error. However, 90% of the time OpenCFU properly separates all six drawn ROIs in the summary. I was 'stress' testing the software yesterday by having it re-analyze images that were rotated at 0, 10, 45, and 100 degrees (for comparison of reported CFUs); I couldn't make the ROI bug consistently occur then either. I will update once I figure out what I must be doing in order to cause it to malfunction (?). Fortunately it seems to be mostly working with my real data. |
Should show CFU/ROI in the GUI and summary result file somehow.
The text was updated successfully, but these errors were encountered: