RefineSplat removes ambiguous distractors from 3DGS via entropy-aware adaptive masking and density control, releasing an 18-scene Ambiguous wild dataset and reporting SOTA metrics on multiple wild benchmarks.
arXiv preprint arXiv:2404.06109 , year=
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2years
2026 2representative citing papers
Two-view accumulation per optimizer step is the dominant training lever for hybrid-capture 3DGS, explained by a variance-decomposition framework showing within-regime gradient variance dominates over between-regime variance.
citing papers explorer
-
Rectifying Mask via Entropy for Distractor-Free 3DGS in Ambiguous Scenarios
RefineSplat removes ambiguous distractors from 3DGS via entropy-aware adaptive masking and density control, releasing an 18-scene Ambiguous wild dataset and reporting SOTA metrics on multiple wild benchmarks.
-
Two-View Accumulation as the Primary Training Lever for Hybrid-Capture Gaussian Splatting: A Variance-Decomposition View of When Gradient Surgery Helps
Two-view accumulation per optimizer step is the dominant training lever for hybrid-capture 3DGS, explained by a variance-decomposition framework showing within-regime gradient variance dominates over between-regime variance.