This is a preview of multi modal neural network input to the aurivus AI. In English: Our AI can read 2D vector graphics and 3D point clouds at the same time and links objects of both together.
Click on a 2D object and find the 3D object in a point cloud. Works also the other way round.
What excites us most in this experiment: you could say the AI can understand two completely different representations of the same thing and finds the link.
Use-case: Add information from 2D plans to 3D point cloud objects to improve 3D model creation or to run calculations directly on the point cloud.
You could think of more use caes, e.g., thinking of a PnID linked to a 3D model or a point cloud.
Video (cookies required):