Relaxing Accurate Initialization Constraint for 3D Gaussian Splatting

3D Gaussian splatting (3DGS) has recently demonstrated impressive capabilities in real-time novel view synthesis and 3D reconstruction. However, 3DGS heavily depends on the accurate initialization derived from Structure-from-Motion (SfM) methods. When the quality of the initial point cloud deteriora...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Jung, Jaewoo, Han, Jisang, An, Honggyu, Kang, Jiwon, Park, Seonghoon, Kim, Seungryong
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Jung, Jaewoo
Han, Jisang
An, Honggyu
Kang, Jiwon
Park, Seonghoon
Kim, Seungryong
description 3D Gaussian splatting (3DGS) has recently demonstrated impressive capabilities in real-time novel view synthesis and 3D reconstruction. However, 3DGS heavily depends on the accurate initialization derived from Structure-from-Motion (SfM) methods. When the quality of the initial point cloud deteriorates, such as in the presence of noise or when using randomly initialized point cloud, 3DGS often undergoes large performance drops. To address this limitation, we propose a novel optimization strategy dubbed RAIN-GS (Relaing Accurate Initialization Constraint for 3D Gaussian Splatting). Our approach is based on an in-depth analysis of the original 3DGS optimization scheme and the analysis of the SfM initialization in the frequency domain. Leveraging simple modifications based on our analyses, RAIN-GS successfully trains 3D Gaussians from sub-optimal point cloud (e.g., randomly initialized point cloud), effectively relaxing the need for accurate initialization. We demonstrate the efficacy of our strategy through quantitative and qualitative comparisons on multiple datasets, where RAIN-GS trained with random point cloud achieves performance on-par with or even better than 3DGS trained with accurate SfM point cloud. Our project page and code can be found at https://ku-cvlab.github.io/RAIN-GS.
doi_str_mv 10.48550/arxiv.2403.09413
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2403_09413</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2403_09413</sourcerecordid><originalsourceid>FETCH-LOGICAL-a673-d2b388a8e3487f6db89b361c2e1cd86efa5244b72bd3cd3b340fc0f0abf908933</originalsourceid><addsrcrecordid>eNotz71OwzAUhmEvDKjlApjwDSQ4Pk7ijFUopVKlStA9Ov5DR0qdynFR4eqBwvRN7yc9jN1XolS6rsUjpgt9lFIJKEWnKrhl61c_4oXiO19Ze06YPd9GyoQjfWGmKfJ-inNOSDHzMCUOT3yD53kmjPztNGLOP_GS3QQcZ3_3vwt2eF4f-pdit99s-9WuwKaFwkkDWqP2oHQbGmd0Z6CprPSVdbrxAWuplGmlcWAdGFAiWBEEmtAJ3QEs2MPf7dUxnBIdMX0Ov57h6oFv4OVGMA</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Relaxing Accurate Initialization Constraint for 3D Gaussian Splatting</title><source>arXiv.org</source><creator>Jung, Jaewoo ; Han, Jisang ; An, Honggyu ; Kang, Jiwon ; Park, Seonghoon ; Kim, Seungryong</creator><creatorcontrib>Jung, Jaewoo ; Han, Jisang ; An, Honggyu ; Kang, Jiwon ; Park, Seonghoon ; Kim, Seungryong</creatorcontrib><description>3D Gaussian splatting (3DGS) has recently demonstrated impressive capabilities in real-time novel view synthesis and 3D reconstruction. However, 3DGS heavily depends on the accurate initialization derived from Structure-from-Motion (SfM) methods. When the quality of the initial point cloud deteriorates, such as in the presence of noise or when using randomly initialized point cloud, 3DGS often undergoes large performance drops. To address this limitation, we propose a novel optimization strategy dubbed RAIN-GS (Relaing Accurate Initialization Constraint for 3D Gaussian Splatting). Our approach is based on an in-depth analysis of the original 3DGS optimization scheme and the analysis of the SfM initialization in the frequency domain. Leveraging simple modifications based on our analyses, RAIN-GS successfully trains 3D Gaussians from sub-optimal point cloud (e.g., randomly initialized point cloud), effectively relaxing the need for accurate initialization. We demonstrate the efficacy of our strategy through quantitative and qualitative comparisons on multiple datasets, where RAIN-GS trained with random point cloud achieves performance on-par with or even better than 3DGS trained with accurate SfM point cloud. Our project page and code can be found at https://ku-cvlab.github.io/RAIN-GS.</description><identifier>DOI: 10.48550/arxiv.2403.09413</identifier><language>eng</language><subject>Computer Science - Computer Vision and Pattern Recognition</subject><creationdate>2024-03</creationdate><rights>http://creativecommons.org/licenses/by/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2403.09413$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2403.09413$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Jung, Jaewoo</creatorcontrib><creatorcontrib>Han, Jisang</creatorcontrib><creatorcontrib>An, Honggyu</creatorcontrib><creatorcontrib>Kang, Jiwon</creatorcontrib><creatorcontrib>Park, Seonghoon</creatorcontrib><creatorcontrib>Kim, Seungryong</creatorcontrib><title>Relaxing Accurate Initialization Constraint for 3D Gaussian Splatting</title><description>3D Gaussian splatting (3DGS) has recently demonstrated impressive capabilities in real-time novel view synthesis and 3D reconstruction. However, 3DGS heavily depends on the accurate initialization derived from Structure-from-Motion (SfM) methods. When the quality of the initial point cloud deteriorates, such as in the presence of noise or when using randomly initialized point cloud, 3DGS often undergoes large performance drops. To address this limitation, we propose a novel optimization strategy dubbed RAIN-GS (Relaing Accurate Initialization Constraint for 3D Gaussian Splatting). Our approach is based on an in-depth analysis of the original 3DGS optimization scheme and the analysis of the SfM initialization in the frequency domain. Leveraging simple modifications based on our analyses, RAIN-GS successfully trains 3D Gaussians from sub-optimal point cloud (e.g., randomly initialized point cloud), effectively relaxing the need for accurate initialization. We demonstrate the efficacy of our strategy through quantitative and qualitative comparisons on multiple datasets, where RAIN-GS trained with random point cloud achieves performance on-par with or even better than 3DGS trained with accurate SfM point cloud. Our project page and code can be found at https://ku-cvlab.github.io/RAIN-GS.</description><subject>Computer Science - Computer Vision and Pattern Recognition</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotz71OwzAUhmEvDKjlApjwDSQ4Pk7ijFUopVKlStA9Ov5DR0qdynFR4eqBwvRN7yc9jN1XolS6rsUjpgt9lFIJKEWnKrhl61c_4oXiO19Ze06YPd9GyoQjfWGmKfJ-inNOSDHzMCUOT3yD53kmjPztNGLOP_GS3QQcZ3_3vwt2eF4f-pdit99s-9WuwKaFwkkDWqP2oHQbGmd0Z6CprPSVdbrxAWuplGmlcWAdGFAiWBEEmtAJ3QEs2MPf7dUxnBIdMX0Ov57h6oFv4OVGMA</recordid><startdate>20240314</startdate><enddate>20240314</enddate><creator>Jung, Jaewoo</creator><creator>Han, Jisang</creator><creator>An, Honggyu</creator><creator>Kang, Jiwon</creator><creator>Park, Seonghoon</creator><creator>Kim, Seungryong</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20240314</creationdate><title>Relaxing Accurate Initialization Constraint for 3D Gaussian Splatting</title><author>Jung, Jaewoo ; Han, Jisang ; An, Honggyu ; Kang, Jiwon ; Park, Seonghoon ; Kim, Seungryong</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a673-d2b388a8e3487f6db89b361c2e1cd86efa5244b72bd3cd3b340fc0f0abf908933</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Computer Science - Computer Vision and Pattern Recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Jung, Jaewoo</creatorcontrib><creatorcontrib>Han, Jisang</creatorcontrib><creatorcontrib>An, Honggyu</creatorcontrib><creatorcontrib>Kang, Jiwon</creatorcontrib><creatorcontrib>Park, Seonghoon</creatorcontrib><creatorcontrib>Kim, Seungryong</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Jung, Jaewoo</au><au>Han, Jisang</au><au>An, Honggyu</au><au>Kang, Jiwon</au><au>Park, Seonghoon</au><au>Kim, Seungryong</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Relaxing Accurate Initialization Constraint for 3D Gaussian Splatting</atitle><date>2024-03-14</date><risdate>2024</risdate><abstract>3D Gaussian splatting (3DGS) has recently demonstrated impressive capabilities in real-time novel view synthesis and 3D reconstruction. However, 3DGS heavily depends on the accurate initialization derived from Structure-from-Motion (SfM) methods. When the quality of the initial point cloud deteriorates, such as in the presence of noise or when using randomly initialized point cloud, 3DGS often undergoes large performance drops. To address this limitation, we propose a novel optimization strategy dubbed RAIN-GS (Relaing Accurate Initialization Constraint for 3D Gaussian Splatting). Our approach is based on an in-depth analysis of the original 3DGS optimization scheme and the analysis of the SfM initialization in the frequency domain. Leveraging simple modifications based on our analyses, RAIN-GS successfully trains 3D Gaussians from sub-optimal point cloud (e.g., randomly initialized point cloud), effectively relaxing the need for accurate initialization. We demonstrate the efficacy of our strategy through quantitative and qualitative comparisons on multiple datasets, where RAIN-GS trained with random point cloud achieves performance on-par with or even better than 3DGS trained with accurate SfM point cloud. Our project page and code can be found at https://ku-cvlab.github.io/RAIN-GS.</abstract><doi>10.48550/arxiv.2403.09413</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2403.09413
ispartof
issn
language eng
recordid cdi_arxiv_primary_2403_09413
source arXiv.org
subjects Computer Science - Computer Vision and Pattern Recognition
title Relaxing Accurate Initialization Constraint for 3D Gaussian Splatting
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-04T08%3A18%3A49IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Relaxing%20Accurate%20Initialization%20Constraint%20for%203D%20Gaussian%20Splatting&rft.au=Jung,%20Jaewoo&rft.date=2024-03-14&rft_id=info:doi/10.48550/arxiv.2403.09413&rft_dat=%3Carxiv_GOX%3E2403_09413%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true