From pictures to real time video object detection on Jetson Nano

I am trying to ramp up a project of model train real time detection on Jetson Nano 4GB.
The idea is to identify a certain number of locomotives (3 as for now). The Jetson Nano will identify each locomotive with itโ€™s coordinate and direction.
As a result, the locomotives will be controlled (speed, direction etc.).
My plan is to use Yolov5, train the 3 locomotives model, and implement on the Jetson Nano.
I would appreciate feedback if my selections are reasonable, or is there a better approach?
Also, Is there a clear guide that show all the way from pictures to a code running on Jetson Nano?
About the Jetson: I chose it because I have one device. Is that a good choice? Or should I go for another?