A sparse-view 3D Gaussian Splatting method that learns per-Gaussian uncertainty to modulate opacity and apply differentiable soft dropout, reporting small PSNR gains over three baselines.
T.; Mildenhall, B.; Verbin, D.; Srinivasan, P
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
UGOD: Uncertainty-Guided Differentiable Opacity and Soft Dropout for Enhanced Sparse-View 3DGS
A sparse-view 3D Gaussian Splatting method that learns per-Gaussian uncertainty to modulate opacity and apply differentiable soft dropout, reporting small PSNR gains over three baselines.