PDF-GS: Progressive Distractor Filtering
for Robust 3D Gaussian Splatting

CVPR 2026 Findings
Kangmin Seo, MinKyu Lee, Tae-Young Kim, ByeongCheol Lee, JoonSeoung An, Jae-Pil Heo
Sungkyunkwan University (SKKU)
arXiv Poster Code
TL;DR — PDF-GS re-purposes 3DGS training as a progressive distractor filter, amplifying its inherent filtering property for robust, distractor-free reconstruction.
3DGS
RobustSplat
Ours

Method Overview

PDF-GS method overview

Abstract

Abstract figure

Recent advances in 3D Gaussian Splatting (3DGS) have enabled impressive real-time photorealistic rendering. However, conventional training pipelines inherently assume full multi-view consistency among input images, which makes them sensitive to distractors that violate this assumption and cause visual artifacts. In this work, we revisit an underexplored aspect of 3DGS: its inherent ability to suppress inconsistent signals. Building on this insight, we propose PDF-GS (Progressive Distractor Filtering for Robust 3D Gaussian Splatting), a framework that amplifies this self-filtering property through a progressive multi-phase optimization. The progressive filtering phases gradually remove distractors by exploiting discrepancy cues, while the following reconstruction phase restores fine-grained, view-consistent details from the purified Gaussian representation. Through this iterative refinement, PDF-GS achieves robust, high-fidelity, and distractor-free reconstructions, consistently outperforming baselines across diverse datasets and challenging real-world conditions. Moreover, our approach is lightweight and easily adaptable to existing 3DGS frameworks, requiring no architectural changes or additional inference overhead, leading to a new state-of-the-art performance.

Motivation

By aggregating multi-view supervision, 3DGS often causes view-inconsistent distractors to disappear or become blurred.

Motivation figure 1

This property allows us to re-purpose 3DGS itself as a distractor filter.

Motivation figure 2

Results

Qualitative comparison
Qualitative comparison
Progressive filtering masks
Progressive distractor mask refinement
NeRF On-the-go benchmark
NeRF On-the-go dataset
RobustNeRF benchmark
RobustNeRF dataset

Citation

@InProceedings{Seo_2026_CVPR,
    author    = {Seo, Kangmin and Lee, MinKyu and Kim, Tae-Young and Lee, ByeongCheol and An, JoonSeoung and Heo, Jae-Pil},
    title     = {PDF-GS: Progressive Distractor Filtering for Robust 3D Gaussian Splatting},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Findings},
    month     = {June},
    year      = {2026},
    pages     = {468-477}
}