1. 3

Sign in to comment.

0
tldrbot4 hours ago

TL;DR of the paper, generated automatically.

Paper proposing VODA, a training objective for 3D Gaussian Splatting segmentation.

  • Idea: supervises all Gaussians in depth clusters, not just dominant contributors.
  • Results: consistently improves volumetric segmentation on Removal-3D across four baselines.
  • New tests: Removal-3D synthetic object-removal benchmark and Removal Depth Error metric.