Representing the bounding box in object detection as a triangle or a circle

Standard object detection(be it 2D or 3D) tends to represent the bounding box as a rectangle or a cuboid(for 3D).
Wanted to know if it’s possible to enclose the detected object by any other shape say a triangle or a circle,and if yes then how should I go about it.

I know that standard box coords are in the following form

(x1,y1) (x2,y2) where (x1,y1) represent the top left coordinates and (x2,y2) represent the bottom right.

sure.

exact values depend on how and what you’ve detected.

what are you detecting and how?

what’s the specific issue you have? calculating a circle’s center and radius?

let’s say that I have detected a vehicle(car) as shown below


Here instead of having a rectangular bounding box,I was wondering on how to use these coordinates to draw a circle enclosing the car.
Another image(better result)

as for the circle 's center and radius sure let’s go with that.Sorry but this is an idea I am experimenting with.
The box coordinates are as follows

387.58 182.51 424.55 203.61

which circle would you like? something different from this?

checkout the minEnclosingXXX() functions here:

https://docs.opencv.org/4.x/d3/dc0/group__imgproc__shape.html#ga8ce13c24081bbc7151e9326f412190f1

I think the outermost circle seems like a good approximation for the bounding box.I am assuming we use the top left and bottom right coordinates to find the mid point,and the radius and then we use that information to obtain the circle?

thanks for the suggestion berak