{"paper":{"title":"2D-SuGaR: Surface-Aware Gaussian Splatting for Geometrically Accurate Mesh Reconstruction","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"Monocular depth and normal priors guide 2D Gaussian Splatting to produce more accurate surface meshes from multi-view images.","cross_cats":["cs.GR"],"primary_cat":"cs.CV","authors_text":"Divyam Sheth, Jinjoo Ha, Justus Thies, Mirela Ostrek, Prajwal Gupta C. R.","submitted_at":"2026-05-01T11:09:29Z","abstract_excerpt":"3D Gaussian Splatting (3DGS) has emerged as a powerful technique for generating photorealistic renderings of a scene in real-time. However, the volumetric nature of 3DGS limits its ability to accurately capture surface geometry. To address this, 2D Gaussian Splatting (2DGS) was proposed to enable view-consistent and geometrically accurate surface reconstruction from multi-view images. However, 2DGS can be sensitive to the initialization of the Gaussian primitives. Reliance on Structure-from-Motion (SfM) initializations, which can produce poor estimates on challenging image sets, may lead to su"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We evaluate our method on the DTU dataset, where it achieves state-of-the-art results in mesh reconstruction while preserving high-quality novel view synthesis.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"Monocular depth and normal priors are sufficiently accurate to guide Gaussian initialization and enable effective pruning of degenerate primitives, particularly when SfM-based initializations are poor.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"2D-SuGaR improves 2D Gaussian Splatting with monocular priors and targeted initialization/pruning to achieve state-of-the-art mesh reconstruction on the DTU dataset while retaining high-quality novel view synthesis.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Monocular depth and normal priors guide 2D Gaussian Splatting to produce more accurate surface meshes from multi-view images.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"96722f2ae0529e85e48fe964918b6a4b7e35affb91c4ca3cd6f8cc6ef36f1e60"},"source":{"id":"2605.00569","kind":"arxiv","version":1},"verdict":{"id":"a0be5f34-e0a0-47be-a52d-e2b5d83b992f","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-09T19:15:37.948851Z","strongest_claim":"We evaluate our method on the DTU dataset, where it achieves state-of-the-art results in mesh reconstruction while preserving high-quality novel view synthesis.","one_line_summary":"2D-SuGaR improves 2D Gaussian Splatting with monocular priors and targeted initialization/pruning to achieve state-of-the-art mesh reconstruction on the DTU dataset while retaining high-quality novel view synthesis.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"Monocular depth and normal priors are sufficiently accurate to guide Gaussian initialization and enable effective pruning of degenerate primitives, particularly when SfM-based initializations are poor.","pith_extraction_headline":"Monocular depth and normal priors guide 2D Gaussian Splatting to produce more accurate surface meshes from multi-view images."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.00569/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-20T19:39:33.824087Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T18:02:42.893974Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"1c0a329cacaede23c4037c88147abdc64b7667dc173088338cfe040a7156d11a"},"references":{"count":12,"sample":[{"doi":"","year":null,"title":"IEEE Transactions on Pattern Analysis and Machine Intelligence , year=","work_id":"75f27b83-0eb8-4496-9dc1-0562b39cfee3","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"3D Gaussian splatting for real-time radiance field rendering. , author=. ACM Trans. Graph. , volume=","work_id":"4bb79b8b-55fc-4b2f-9748-1242532b4734","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"ACM SIGGRAPH 2024 conference papers , pages=","work_id":"6fb715c8-d0b0-4ce3-bb9c-1753192a6c0e","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=","work_id":"96c01215-8c1f-4807-93d6-4ddd7db7090f","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis , author=. 2020 , booktitle=","work_id":"77b3597d-2b5f-49b7-8502-8e0500842e76","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":12,"snapshot_sha256":"937bf056dfa835f1a204defa71867f3ba6ffe6e4e3ed89c609c760cfe8021344","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}