Spatial Relations
From TinyCog
This subsystems converts a Scene Graph with 3D Object information from the 3D Reconstruction subsystem into a symbolic representation.
- Input is a set of successfully classified Objects with their position and other parameters.
- Output is a Description Logic "ABox" with a list of objects and DL roles describing their spatial relationships.
Figure 1: A man stands in front of a table.
table1 :: table and
at_left : human1 and
on_top : floor1.
human1 :: human and
at_right : table1 and
in_front_of : table1 and
on_top : floor1.
Figure 2: A Description Logic representation of the image above.
Implementation
TinyCog 0.0.2 implements basic 2D spatial analysis of objects in the Hunter Domain. Deer close to hunters and deer close to grazing grounds are recognized and encoded similar to the example above.
