Search papers, labs, and topics across Lattice.
This paper introduces PDF-GS, a novel framework that enhances the robustness of 3D Gaussian Splatting (3DGS) against distractors by exploiting its inherent self-filtering capability. PDF-GS employs a progressive multi-phase optimization strategy, iteratively removing distractors based on discrepancy cues and subsequently refining the Gaussian representation to restore view-consistent details. Experiments demonstrate that PDF-GS consistently outperforms existing methods in reconstructing high-fidelity, distractor-free 3D scenes across various datasets without architectural changes or inference overhead.
3D Gaussian Splatting can self-filter inconsistent signals, and this paper shows how to amplify that property to achieve state-of-the-art robustness against distractors without sacrificing performance.
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. The code is publicly available at https://github.com/kangrnin/PDF-GS.
