Custom Pattern Recognition - Python

I'm thinking a diff. approach for custom object detection (such as bud,stem,flower, and leaves formation of a distance of at least 1-2 meters) for the past few weeks.

But, I don't know if this makes any sense or if it's even possible (I've already read and watch dozens of tutorials about pattern recognition but almost all of them only works if the pattern is not too complex like traffic signs, letters and numbers).

Is it possible with opencv (or other framework as long as it's compatible with python) to get pattern of leaves formation (not just a leaf but the general formation of the leaves on the top part of a plant) from a distance of 1-2 meters(that's the aprox. distance of my device from the target object)?

The idea is to get a pattern template in the form of connected lines or dots (whichever is feasible) based on the formation of plant leaves then compare it on a cluttered image and see if there's a match on the pattern somewhere in the image.

I've tried using morphography (not sure with the spelling) and zhang-suen algo to generate a pattern (something like a skeleton) but i failed because of the distance and lots of noise in the background. I'm considering this method (pattern matching) since the other algo. that I've tried like viola-jones, feature matching, etc. didn't work so well coz the disease infected plant that I'm trying to detect and recognize is almost the same in visual appearance with the healthy ones (only the leaves formation is noticeably diff.).

When I say leaves formation, it’s the broom-like leaf proliferation at the top of the plant.

I’m hoping that someone could give me direction on this as I’m already on a dead end.