haars cascades not that great ?

As far as I remember Haar cascades are outperformed by modern CNNs. Their only advantage was (if I remember right) lesser computational complexity than CNNs, but with modern chips with hardware acceleration for CNNs it's not really a problem any more.

Yea now what are some of these since Google is being kind of not helpful giving me unhelpful results

@MadMadMadMrMim just search for "deep learning {use case} network", get the trained model from the tensorflow site or GitHub or whatever, and run it. You'll probably need to install Tensorflow and/or PyTorch to run them. Training one of these nets with your own data is a tough job though. They're mostly supervised learning, so you'll need a suitably prepared dataset. But they come pretrained for common use cases. For object detection/localization, check out YOLO v3 and MobileNet v2, among others.

You could also search for tutorials on using whichever networks you find. Ultimately though, computer vision is a complex job, no way around that.

@RememberMe well it’s more about it being a definitive set of instructions for setup since the goal would seem to be that so you could move on to applying the exhausting crap

But what if I want to train my own object models is the problem I’m reaching

Like more than faces
I see examples but that doesn’t seem to add up

But yeah I will look at your suggestions

@RememberMe thanks again