Comparison of implementations of a random solution in C ++

I am currently converting some R code into C ++ code, and I need a “good” C ++ implementation of Random Decision Forest.

So far, I have discovered three large implementations (tmva, alglib, and openCv), some “small / obsolete” ones (like librf), and I need to choose one of them:

Do you guys have any good / bad impressions and / or some recommendations regarding these libraries (or maybe some others)? For example, ease of use, mobility, memory usage, speed, readability of error messages, errors (?), Comments on implementation options, etc.

If you want to know, I work with Visual Studio, but my code (and should remain) is compatible with Linux. Speed ​​and memory usage are very important to me, as I will compute and store a large number of random forests in my memory. The code I am developing is a machine learning algorithm for symbolic time sequences.

Thanks in advance,