Question about the Object-level Data Generation #8
Description
In the "Object-Level Dataset Generation" section, you mentioned that "for objects with complex kinematic structures, we procedurally generate a large number of examples of these objects." I would like to know which model or approach you used to generate these articulated objects. I have found that existing methods such as Procthor[11], Holodeck[49], and infiniteGen[17] primarily support scene-level procedural generation. Could you please share more information about this process?