WildGaussians: 3D Gaussian Splatting in the Wild

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F24%3A00383949" target="_blank" >RIV/68407700:21230/24:00383949 - isvavai.cz</a>

Nalezeny alternativní kódy

RIV/68407700:21730/24:00383949

Výsledek na webu

<a href="https://papers.nips.cc/paper_files/paper/2024/file/25c0fe7b157821dd3140727dc07461da-Paper-Conference.pdf" target="_blank" >https://papers.nips.cc/paper_files/paper/2024/file/25c0fe7b157821dd3140727dc07461da-Paper-Conference.pdf</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

WildGaussians: 3D Gaussian Splatting in the Wild

Popis výsledku v původním jazyce

While the field of 3D scene reconstruction is dominated by NeRFs due to their pho torealistic quality, 3D Gaussian Splatting (3DGS) has recently emerged, offering similar quality with real-time rendering speeds. However, both methods primarily excel with well-controlled 3D scenes, while in-the-wild data– characterized by oc clusions, dynamic objects, and varying illumination– remains challenging. NeRFs can adapt to such conditions easily through per-image embedding vectors, but 3DGSstruggles due to its explicit representation and lack of shared parameters. To address this, we introduce WildGaussians, a novel approach to handle occlusions and appearance changes with 3DGS. By leveraging robust DINO features and integrating an appearance modeling module within 3DGS, our method achieves state-of-the-art results. We demonstrate that WildGaussians matches the real-time rendering speed of 3DGS while surpassing both 3DGS and NeRF baselines in handling in-the-wild data, all within a simple architectural framework.

Název v anglickém jazyce

WildGaussians: 3D Gaussian Splatting in the Wild

Popis výsledku anglicky

While the field of 3D scene reconstruction is dominated by NeRFs due to their pho torealistic quality, 3D Gaussian Splatting (3DGS) has recently emerged, offering similar quality with real-time rendering speeds. However, both methods primarily excel with well-controlled 3D scenes, while in-the-wild data– characterized by oc clusions, dynamic objects, and varying illumination– remains challenging. NeRFs can adapt to such conditions easily through per-image embedding vectors, but 3DGSstruggles due to its explicit representation and lack of shared parameters. To address this, we introduce WildGaussians, a novel approach to handle occlusions and appearance changes with 3DGS. By leveraging robust DINO features and integrating an appearance modeling module within 3DGS, our method achieves state-of-the-art results. We demonstrate that WildGaussians matches the real-time rendering speed of 3DGS while surpassing both 3DGS and NeRF baselines in handling in-the-wild data, all within a simple architectural framework.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

<a href="/cs/project/GX23-07973X" target="_blank" >GX23-07973X: Sjednocená Reprezentace 3D Map</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Advances in Neural Information Processing Systems 37 (NeurIPS 2024)

ISBN

9798331314385

ISSN

1049-5258

e-ISSN

1049-5258

Počet stran výsledku

18

Strana od-do

—

Název nakladatele

Neural Information Processing Systems Foundation, Inc.

Místo vydání

—

Místo konání akce

Vancouver

Datum konání akce

10. 12. 2024

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

001633259400234