OpenCV is intended to be useful for computer vision researchers to test new functionality easily and also for developers to put together CV based functionality easily. I think the new C++ interface is pretty neat. Although I do agree that it is becoming humongous, that is a sign of maturity rather than neglect. As a computer vision developer, I believe that choice of algorithms, and ability to use low level functionality is key to developing robust computer vision applications. Mostly because there is no one size fits all. Probably you are aiming for fewer use cases than what is possible with OpenCV. In that case, it makes sense.