How to make a custom YOLO package

edit

I have already made a raspberry pi camera show images thru rviz. I have this neat node package. How can I launch yolo thru ROS if it isn't a package? Really confused here. And one more thing, I'm not really familiar with using yolo.

You have to make yolo a ROS package, I don't think there is another solution to that.

How to do that exactly? Any tutorials and what nots you can share?

Unless I'm misunderstanding something, there appears to be no reason to create yet another ROS YOLO pkg, as there are quite a few available already. I'd really advise to take a look at those, as they potentially save you a lot of work.

As mentioned in the question that they are customizing yolo, I am assuming they are making some specific changes for their case which I am not sure how easy can be to add to an existing package. If only making their own trained model then it should be fairly easy to use existing packages.

The OP will have to clarify this. I've seen people claim they are creating a 'custom deep learning system', but in the end were only loading an alternative model.

Even if the OP really needs a different network or some other customisation, starting with existing pkgs is probably a good thing.

I'm not really sure how the other team is "customizing" YOLO. But if we are to add only trained models, how can we add this to an existing ros yolo like this one?

I'm not really sure how the other team is "customizing" YOLO

then personally I would first try to get some info from them. Seeing as your task appears to be to integrate the results of the 'other team', wouldn't it be good to know what they are actually doing?