High-Fidelity and Generalizable Neural Surface Reconstruction with Sparse Feature Volumes Article Swipe
Aoxiang Fan
,
Corentin Dumery
,
Nicolas Talabot
,
Hieu Van Le
,
Pascal Fua
·
YOU?
·
· 2025
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2507.05952
YOU?
·
· 2025
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2507.05952
Generalizable neural surface reconstruction has become a compelling technique to reconstruct from few images without per-scene optimization, where dense 3D feature volume has proven effective as a global representation of scenes. However, the dense representation does not scale well to increasing voxel resolutions, severely limiting the reconstruction quality. We thus present a sparse representation method, that maximizes memory efficiency and enables significantly higher resolution reconstructions on standard hardware. We implement this through a two-stage approach: First training a network to predict voxel occupancies from posed images and associated depth maps, then computing features and performing volume rendering only in voxels with sufficiently high occupancy estimates. To support this sparse representation, we developed custom algorithms for efficient sampling, feature aggregation, and querying from sparse volumes-overcoming the dense-volume assumptions inherent in existing works. Experiments on public datasets demonstrate that our approach reduces storage requirements by more than 50 times without performance degradation, enabling reconstructions at $512^3$ resolution compared to the typical $128^3$ on similar hardware, and achieving superior reconstruction accuracy over current state-of-the-art methods.
Related Topics
Concepts
No concepts available.
Metadata
- Type
- preprint
- Language
- en
- Landing Page
- http://arxiv.org/abs/2507.05952
- https://arxiv.org/pdf/2507.05952
- OA Status
- green
- OpenAlex ID
- https://openalex.org/W4416061755
All OpenAlex metadata
Raw OpenAlex JSON
- OpenAlex ID
-
https://openalex.org/W4416061755Canonical identifier for this work in OpenAlex
- DOI
-
https://doi.org/10.48550/arxiv.2507.05952Digital Object Identifier
- Title
-
High-Fidelity and Generalizable Neural Surface Reconstruction with Sparse Feature VolumesWork title
- Type
-
preprintOpenAlex work type
- Language
-
enPrimary language
- Publication year
-
2025Year of publication
- Publication date
-
2025-07-08Full publication date if available
- Authors
-
Aoxiang Fan, Corentin Dumery, Nicolas Talabot, Hieu Van Le, Pascal FuaList of authors in order
- Landing page
-
https://arxiv.org/abs/2507.05952Publisher landing page
- PDF URL
-
https://arxiv.org/pdf/2507.05952Direct link to full text PDF
- Open access
-
YesWhether a free full text is available
- OA status
-
greenOpen access status per OpenAlex
- OA URL
-
https://arxiv.org/pdf/2507.05952Direct OA link when available
- Cited by
-
0Total citation count in OpenAlex
Full payload
| id | https://openalex.org/W4416061755 |
|---|---|
| doi | https://doi.org/10.48550/arxiv.2507.05952 |
| ids.doi | https://doi.org/10.48550/arxiv.2507.05952 |
| ids.openalex | https://openalex.org/W4416061755 |
| fwci | |
| type | preprint |
| title | High-Fidelity and Generalizable Neural Surface Reconstruction with Sparse Feature Volumes |
| biblio.issue | |
| biblio.volume | |
| biblio.last_page | |
| biblio.first_page | |
| is_xpac | False |
| apc_list | |
| apc_paid | |
| language | en |
| locations[0].id | pmh:oai:arXiv.org:2507.05952 |
| locations[0].is_oa | True |
| locations[0].source.id | https://openalex.org/S4306400194 |
| locations[0].source.issn | |
| locations[0].source.type | repository |
| locations[0].source.is_oa | True |
| locations[0].source.issn_l | |
| locations[0].source.is_core | False |
| locations[0].source.is_in_doaj | False |
| locations[0].source.display_name | arXiv (Cornell University) |
| locations[0].source.host_organization | https://openalex.org/I205783295 |
| locations[0].source.host_organization_name | Cornell University |
| locations[0].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[0].license | |
| locations[0].pdf_url | https://arxiv.org/pdf/2507.05952 |
| locations[0].version | submittedVersion |
| locations[0].raw_type | text |
| locations[0].license_id | |
| locations[0].is_accepted | False |
| locations[0].is_published | False |
| locations[0].raw_source_name | |
| locations[0].landing_page_url | http://arxiv.org/abs/2507.05952 |
| locations[1].id | doi:10.48550/arxiv.2507.05952 |
| locations[1].is_oa | True |
| locations[1].source.id | https://openalex.org/S4306400194 |
| locations[1].source.issn | |
| locations[1].source.type | repository |
| locations[1].source.is_oa | True |
| locations[1].source.issn_l | |
| locations[1].source.is_core | False |
| locations[1].source.is_in_doaj | False |
| locations[1].source.display_name | arXiv (Cornell University) |
| locations[1].source.host_organization | https://openalex.org/I205783295 |
| locations[1].source.host_organization_name | Cornell University |
| locations[1].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[1].license | cc-by |
| locations[1].pdf_url | |
| locations[1].version | |
| locations[1].raw_type | article |
| locations[1].license_id | https://openalex.org/licenses/cc-by |
| locations[1].is_accepted | False |
| locations[1].is_published | |
| locations[1].raw_source_name | |
| locations[1].landing_page_url | https://doi.org/10.48550/arxiv.2507.05952 |
| indexed_in | arxiv, datacite |
| authorships[0].author.id | https://openalex.org/A5043378752 |
| authorships[0].author.orcid | https://orcid.org/0000-0002-2877-9795 |
| authorships[0].author.display_name | Aoxiang Fan |
| authorships[0].author_position | first |
| authorships[0].raw_author_name | Fan, Aoxiang |
| authorships[0].is_corresponding | False |
| authorships[1].author.id | https://openalex.org/A5008484135 |
| authorships[1].author.orcid | https://orcid.org/0000-0001-5314-7979 |
| authorships[1].author.display_name | Corentin Dumery |
| authorships[1].author_position | middle |
| authorships[1].raw_author_name | Dumery, Corentin |
| authorships[1].is_corresponding | False |
| authorships[2].author.id | https://openalex.org/A5064916555 |
| authorships[2].author.orcid | |
| authorships[2].author.display_name | Nicolas Talabot |
| authorships[2].author_position | middle |
| authorships[2].raw_author_name | Talabot, Nicolas |
| authorships[2].is_corresponding | False |
| authorships[3].author.id | https://openalex.org/A5111779321 |
| authorships[3].author.orcid | |
| authorships[3].author.display_name | Hieu Van Le |
| authorships[3].author_position | middle |
| authorships[3].raw_author_name | Le, Hieu |
| authorships[3].is_corresponding | False |
| authorships[4].author.id | https://openalex.org/A5038674741 |
| authorships[4].author.orcid | https://orcid.org/0000-0002-6702-9970 |
| authorships[4].author.display_name | Pascal Fua |
| authorships[4].author_position | last |
| authorships[4].raw_author_name | Fua, Pascal |
| authorships[4].is_corresponding | False |
| has_content.pdf | False |
| has_content.grobid_xml | False |
| is_paratext | False |
| open_access.is_oa | True |
| open_access.oa_url | https://arxiv.org/pdf/2507.05952 |
| open_access.oa_status | green |
| open_access.any_repository_has_fulltext | False |
| created_date | 2025-10-10T00:00:00 |
| display_name | High-Fidelity and Generalizable Neural Surface Reconstruction with Sparse Feature Volumes |
| has_fulltext | False |
| is_retracted | False |
| updated_date | 2025-11-28T10:25:13.201920 |
| primary_topic | |
| cited_by_count | 0 |
| locations_count | 2 |
| best_oa_location.id | pmh:oai:arXiv.org:2507.05952 |
| best_oa_location.is_oa | True |
| best_oa_location.source.id | https://openalex.org/S4306400194 |
| best_oa_location.source.issn | |
| best_oa_location.source.type | repository |
| best_oa_location.source.is_oa | True |
| best_oa_location.source.issn_l | |
| best_oa_location.source.is_core | False |
| best_oa_location.source.is_in_doaj | False |
| best_oa_location.source.display_name | arXiv (Cornell University) |
| best_oa_location.source.host_organization | https://openalex.org/I205783295 |
| best_oa_location.source.host_organization_name | Cornell University |
| best_oa_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| best_oa_location.license | |
| best_oa_location.pdf_url | https://arxiv.org/pdf/2507.05952 |
| best_oa_location.version | submittedVersion |
| best_oa_location.raw_type | text |
| best_oa_location.license_id | |
| best_oa_location.is_accepted | False |
| best_oa_location.is_published | False |
| best_oa_location.raw_source_name | |
| best_oa_location.landing_page_url | http://arxiv.org/abs/2507.05952 |
| primary_location.id | pmh:oai:arXiv.org:2507.05952 |
| primary_location.is_oa | True |
| primary_location.source.id | https://openalex.org/S4306400194 |
| primary_location.source.issn | |
| primary_location.source.type | repository |
| primary_location.source.is_oa | True |
| primary_location.source.issn_l | |
| primary_location.source.is_core | False |
| primary_location.source.is_in_doaj | False |
| primary_location.source.display_name | arXiv (Cornell University) |
| primary_location.source.host_organization | https://openalex.org/I205783295 |
| primary_location.source.host_organization_name | Cornell University |
| primary_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| primary_location.license | |
| primary_location.pdf_url | https://arxiv.org/pdf/2507.05952 |
| primary_location.version | submittedVersion |
| primary_location.raw_type | text |
| primary_location.license_id | |
| primary_location.is_accepted | False |
| primary_location.is_published | False |
| primary_location.raw_source_name | |
| primary_location.landing_page_url | http://arxiv.org/abs/2507.05952 |
| publication_date | 2025-07-08 |
| publication_year | 2025 |
| referenced_works_count | 0 |
| abstract_inverted_index.a | 6, 26, 51, 72, 77 |
| abstract_inverted_index.3D | 19 |
| abstract_inverted_index.50 | 145 |
| abstract_inverted_index.To | 105 |
| abstract_inverted_index.We | 48, 68 |
| abstract_inverted_index.as | 25 |
| abstract_inverted_index.at | 152 |
| abstract_inverted_index.by | 142 |
| abstract_inverted_index.in | 98, 128 |
| abstract_inverted_index.of | 29 |
| abstract_inverted_index.on | 65, 132, 160 |
| abstract_inverted_index.to | 9, 39, 79, 156 |
| abstract_inverted_index.we | 110 |
| abstract_inverted_index.and | 59, 86, 93, 119, 163 |
| abstract_inverted_index.few | 12 |
| abstract_inverted_index.for | 114 |
| abstract_inverted_index.has | 4, 22 |
| abstract_inverted_index.not | 36 |
| abstract_inverted_index.our | 137 |
| abstract_inverted_index.the | 32, 45, 124, 157 |
| abstract_inverted_index.does | 35 |
| abstract_inverted_index.from | 11, 83, 121 |
| abstract_inverted_index.high | 102 |
| abstract_inverted_index.more | 143 |
| abstract_inverted_index.only | 97 |
| abstract_inverted_index.over | 168 |
| abstract_inverted_index.than | 144 |
| abstract_inverted_index.that | 55, 136 |
| abstract_inverted_index.then | 90 |
| abstract_inverted_index.this | 70, 107 |
| abstract_inverted_index.thus | 49 |
| abstract_inverted_index.well | 38 |
| abstract_inverted_index.with | 100 |
| abstract_inverted_index.First | 75 |
| abstract_inverted_index.dense | 18, 33 |
| abstract_inverted_index.depth | 88 |
| abstract_inverted_index.maps, | 89 |
| abstract_inverted_index.posed | 84 |
| abstract_inverted_index.scale | 37 |
| abstract_inverted_index.times | 146 |
| abstract_inverted_index.voxel | 41, 81 |
| abstract_inverted_index.where | 17 |
| abstract_inverted_index.become | 5 |
| abstract_inverted_index.custom | 112 |
| abstract_inverted_index.global | 27 |
| abstract_inverted_index.higher | 62 |
| abstract_inverted_index.images | 13, 85 |
| abstract_inverted_index.memory | 57 |
| abstract_inverted_index.neural | 1 |
| abstract_inverted_index.proven | 23 |
| abstract_inverted_index.public | 133 |
| abstract_inverted_index.sparse | 52, 108, 122 |
| abstract_inverted_index.volume | 21, 95 |
| abstract_inverted_index.voxels | 99 |
| abstract_inverted_index.works. | 130 |
| abstract_inverted_index.$128^3$ | 159 |
| abstract_inverted_index.$512^3$ | 153 |
| abstract_inverted_index.current | 169 |
| abstract_inverted_index.enables | 60 |
| abstract_inverted_index.feature | 20, 117 |
| abstract_inverted_index.method, | 54 |
| abstract_inverted_index.network | 78 |
| abstract_inverted_index.predict | 80 |
| abstract_inverted_index.present | 50 |
| abstract_inverted_index.reduces | 139 |
| abstract_inverted_index.scenes. | 30 |
| abstract_inverted_index.similar | 161 |
| abstract_inverted_index.storage | 140 |
| abstract_inverted_index.support | 106 |
| abstract_inverted_index.surface | 2 |
| abstract_inverted_index.through | 71 |
| abstract_inverted_index.typical | 158 |
| abstract_inverted_index.without | 14, 147 |
| abstract_inverted_index.However, | 31 |
| abstract_inverted_index.accuracy | 167 |
| abstract_inverted_index.approach | 138 |
| abstract_inverted_index.compared | 155 |
| abstract_inverted_index.datasets | 134 |
| abstract_inverted_index.enabling | 150 |
| abstract_inverted_index.existing | 129 |
| abstract_inverted_index.features | 92 |
| abstract_inverted_index.inherent | 127 |
| abstract_inverted_index.limiting | 44 |
| abstract_inverted_index.methods. | 171 |
| abstract_inverted_index.quality. | 47 |
| abstract_inverted_index.querying | 120 |
| abstract_inverted_index.severely | 43 |
| abstract_inverted_index.standard | 66 |
| abstract_inverted_index.superior | 165 |
| abstract_inverted_index.training | 76 |
| abstract_inverted_index.achieving | 164 |
| abstract_inverted_index.approach: | 74 |
| abstract_inverted_index.computing | 91 |
| abstract_inverted_index.developed | 111 |
| abstract_inverted_index.effective | 24 |
| abstract_inverted_index.efficient | 115 |
| abstract_inverted_index.hardware, | 162 |
| abstract_inverted_index.hardware. | 67 |
| abstract_inverted_index.implement | 69 |
| abstract_inverted_index.maximizes | 56 |
| abstract_inverted_index.occupancy | 103 |
| abstract_inverted_index.per-scene | 15 |
| abstract_inverted_index.rendering | 96 |
| abstract_inverted_index.sampling, | 116 |
| abstract_inverted_index.technique | 8 |
| abstract_inverted_index.two-stage | 73 |
| abstract_inverted_index.algorithms | 113 |
| abstract_inverted_index.associated | 87 |
| abstract_inverted_index.compelling | 7 |
| abstract_inverted_index.efficiency | 58 |
| abstract_inverted_index.estimates. | 104 |
| abstract_inverted_index.increasing | 40 |
| abstract_inverted_index.performing | 94 |
| abstract_inverted_index.resolution | 63, 154 |
| abstract_inverted_index.Experiments | 131 |
| abstract_inverted_index.assumptions | 126 |
| abstract_inverted_index.demonstrate | 135 |
| abstract_inverted_index.occupancies | 82 |
| abstract_inverted_index.performance | 148 |
| abstract_inverted_index.reconstruct | 10 |
| abstract_inverted_index.aggregation, | 118 |
| abstract_inverted_index.degradation, | 149 |
| abstract_inverted_index.dense-volume | 125 |
| abstract_inverted_index.requirements | 141 |
| abstract_inverted_index.resolutions, | 42 |
| abstract_inverted_index.sufficiently | 101 |
| abstract_inverted_index.Generalizable | 0 |
| abstract_inverted_index.optimization, | 16 |
| abstract_inverted_index.significantly | 61 |
| abstract_inverted_index.reconstruction | 3, 46, 166 |
| abstract_inverted_index.representation | 28, 34, 53 |
| abstract_inverted_index.reconstructions | 64, 151 |
| abstract_inverted_index.representation, | 109 |
| abstract_inverted_index.state-of-the-art | 170 |
| abstract_inverted_index.volumes-overcoming | 123 |
| cited_by_percentile_year | |
| countries_distinct_count | 0 |
| institutions_distinct_count | 5 |
| citation_normalized_percentile |