Is there any feasible way to use a model created using teachable machine in processing?

I’m in between converting to P5 a huge amount of matrix calculations I made a while ago for processing… because, then, I can use the ml5 library with image classification.
But that’s a bit silly if I can have a simple way to make a processing equivalent library which the minimal requirements to deal with a model downloaded from Image Classification. Any updates on the use of image classification in processing?