Probably plenty of other information out here on the forum.
One thing I would try is something like jit.chromakey to isolate the color you want, then feed that output into a jit.rgb2luma into something like cv.jit.blobs so that you're only doing computer vision on that one color.
You could also do it with a jit.expr or getting fancy with jit.op depending on the color, but you would be best to isolate the color first, and then do blob counting and depending on your skillset, jit.chromakey is probably easiest
I'm not a big expert with max, but with what you told me and some search, i could copy some parts of other patches and build mine. Now i can detect blobs of a determined color and know how many they are.
This is what i have:
But i'm having a problem. If I have different size blobs, the counting could be wrong, 'cause the radius that is ok for signing a small blob will fill a bigger blob full of blob signs.
The cv.jit.thouches, gives me the blobs and the number of blobs, that's nice.
The cv.jit.blobs.bounds, gives me the exact form of the blob. But i can't count the number of blobs.
Can i have both? Or something to solve this problem?