Look Where You Can: Active View Selection for CAD Reconstruction under Occlusion

Key points

• CAD reconstruction methods assume a luxury reality rarely grants: unrestricted visual access to the object, photographed from any desired angle.1
• In this work we consider objects from standard CAD benchmarks embedded in realistic indoor scenes with physically derived visibility constraints over a discrete view sphere.1
• A learned view selector must choose $K$ feasible views for a vision--language model (VLM) that generates executable CadQuery code, scored by geometric fidelity of the executed solid.1
• Because reward arrives only after discrete view selection, autoregressive generation, and CAD-kernel execution, we propose a joint training paradigm in which the view selector and the CAD-generation VLM are trained together against this reward.1

Sources (1)

• [1]Look Where You Can: Active View Selection for CAD Reconstruction under Occlusion
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 01:40 PM

  We introduce SightCAD, a framework for parametric CAD reconstruction that treats view feasibility as a first-class constraint.

  CAD reconstruction methods assume a luxury reality rarely grants: unrestricted visual access to the object, photographed from any desired angle.

Extractive summary: sentences quoted from the sources.