everyone!
i am trying to find distance of an object(apple for example) from my camera(using a single camera) by using depth map created by neural network(like pydnet or MiDaS)
I have two problems:
I need a really good fps
I want to run the code in Embedded(like Jetson Nano) so I don't have a strong computer to calculate fast.
now I am detecting the object(say apple) with Yolo and drawing a rectangle around it. I wanted to know can I have a depth map of the rectangle? I mean instead of calculating depth map of the whole picture just calculate depth map of the rectangle. no matter what neural network is in use, i just want a fast one with a good performance thank you :)
also, it will be helpful if you know any other way to get distance of the object from my camera, just notice that my robot is moving fast and I need a good performance and the rectangle transformation did not work well for me
i have found the answer myself! just crop the object and send it to the model. its not as accurate as whole picture but its looking good.