Window Dimensions API OpenCV Python

An API is to be provided that receives an image file of a window section taken for a construction project. The image comes from the camera of a mobile device (smartphone/tablet) and meets the following requirements:

The window is photographed straight from the front.
The window is rectangular or square.
The image is well-lit.
A reference object (e.g., a green cube) with known dimensions can be placed in the image.
The API detects the object (window or window hole) and generates the outer dimensions. The tolerance should be as low as possible (max 1.0 - 1.5 cm).

The API returns a JSON with the following properties:

width: Width of the window hole
height: Height of the window hole
imagepath: Image path to the picture with the frame and dimensions, so that it can be checked if the section was captured correctly

Only the API needs to be programmed. The image illustrates Step 2 and Step 3.

Ideally, the project will be implemented in Python.