|
shape recognition
Anonymous from United States [2 posts] |
10 year
|
Hi There : Im totally new here but would like to know if RoboRealm will help me do the following:
I have a USB microscope camera that looks at particles on a filter. I have training sets of similar images in the following categories ; Particle A, Particle B, Particle C, Particle D. I have about 100 training images so far in each category. I would then like to scan my microscope slide , count the particles and categorize them into each type bin A B C and D. So I would have something like
Type A = 20 particles
Type B = 5 particles
Type C = 0 particles
Type D = 100 particles
Maybe a break down of the size distribution in each category would be useful.
Is this remotely possible with RoboRealm, the shape recognition module seems very close.
Thanks for any help
|
|
|
|
Steven Gentner from United States [1446 posts] |
10 year
|
Yes, this is possible but the shape recognition module is probably not the right solution. That's meant to identify objects that look very similar as apposed to share similar traits. You probably want the Blob Recognition module instead. Its hard to determine this without any example images. Are you able to post a couple?
STeven.
|
|
|
|
Anonymous from United States [2 posts] |
10 year
|
Thank you Steven: I am going with the shape recognition route for now it seems to be working well just running manually tweaking the settings in the modules. The issue im having with the module is getting all the data out of the routine given my confidence and size filters because the shape_match only outputs the best match in the module. I would like to output sizes and counts for each of my categories. I can see this is possible using the shape array but im struggling using scripts to try and get this data into excel for processing.
i actually sent you a private e-mail about how to go about getting this data out from the shape array.
Also is there any way to get the individual images out for each categorized shape category.
Thanks very much for help - im getting there slowly but surely.
Thanks again
|
|
This Thread