Question Agent with Computer Vision capabilities

I want to creat a Drone Agent that can find car objects in a 3D map( the cars and 3D map are real, produced from a physical DJI Drone).

There are a lot of 3D computer Vision algos out there that can find a car in a 3D env.
How can I implement that vision in the Drone Agent in Unity?
I want the Drone Agent to be capble of detecting car objects and get a reward by doing so.

Thank you,
Itamar.

 

Unity ml-agents has a vision sensor. But its kind of not working for me well. More on this:

https://forum.unity.com/threads/render-to-texture-sensor-eats-up-memory-fast.1377777/
https://github.com/Unity-Technologies/ml-agents/issues/5852

And even worse news:

https://forum.unity.com/threads/is-ml-agents-dead.1297374/page-2#post-8724534