Ask Your Question

Is there any standard data format for 3D reconstruction?

asked 2017-11-14 04:26:39 -0500

When we create a 3D reconstruction, we convert a bitmap picture into some symbolic representation. This description will contain such elements as object names (box, ball, line ...), their numeric attributes (length, height, coordinates ...), symbolic attributes (color ...) ... Is there any widespread format for such representation?

edit retag flag offensive close merge delete


3d reconstruction gives you 3d points (a cloud), that's it.

it does not know anything about objects, structure, or names, all these things would need human annotation.

berak gravatar imageberak ( 2017-11-14 10:00:01 -0500 )edit

A shape of the object is needed already now because this object needs to be detected in the image first. Examples which are present in OpenCV handle specifically chessboard. They will not work with a bottle or some book. In the future, it is desirable to have a ready object at the output. It is not difficult if you can recognize it at the input.

ya_ocv_user gravatar imageya_ocv_user ( 2017-11-14 11:19:05 -0500 )edit

1 answer

Sort by » oldest newest most voted

answered 2017-11-14 07:38:05 -0500

Yijie gravatar image

Maybe you can use Industry Foundation Classes(IFC) for 3D reconstruction of buildings or indoor spaces

edit flag offensive delete link more


Yes, creating the plan of furniture in a room is one possible task. Can you provide a link?

ya_ocv_user gravatar imageya_ocv_user ( 2017-11-14 08:27:15 -0500 )edit

IFC That what is required. It supports Attribute-driven geometry representation among other things.

ya_ocv_user gravatar imageya_ocv_user ( 2017-11-14 19:13:35 -0500 )edit

Question Tools

1 follower


Asked: 2017-11-14 04:26:39 -0500

Seen: 204 times

Last updated: Nov 14 '17