{"paper":{"title":"Contrastive Language-Colored Pointmap Pretraining for Unified 3D Scene Understanding","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"Pretraining a transformer on multi-view colored pointmaps with language contrast produces unified 3D scene representations.","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Junpeng Jing, Krystian Mikolajczyk, Ranran Huang, Weixun Luo, Ye Mao","submitted_at":"2026-04-02T21:54:43Z","abstract_excerpt":"Pretraining 3D encoders by aligning with Contrastive Language Image Pretraining (CLIP) has emerged as a promising direction to learn generalizable representations for 3D scene understanding. In this paper, we propose UniScene3D, a transformer-based encoder that learns unified scene representations from multi-view colored pointmaps, jointly modeling image appearance and geometry. For robust colored pointmap representation learning, we introduce novel cross-view geometric alignment and grounded view alignment to enforce cross-view geometry and semantic consistency. Extensive low-shot and task-sp"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We propose UniScene3D, a transformer-based encoder that learns unified scene representations from multi-view colored pointmaps, jointly modeling image appearance and geometry. ... Extensive low-shot and task-specific fine-tuning evaluations on viewpoint grounding, scene retrieval, scene type classification, and 3D VQA demonstrate our state-of-the-art performance.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the introduced cross-view geometric alignment and grounded view alignment will successfully enforce cross-view geometry and semantic consistency, leading to more generalizable unified representations (assumed without detailed verification of failure modes or data requirements in the abstract).","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"UniScene3D learns unified 3D scene representations from colored pointmaps using contrastive CLIP pretraining plus cross-view geometric and grounded view alignments, achieving state-of-the-art results on viewpoint grounding, scene retrieval, classification, and 3D VQA.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Pretraining a transformer on multi-view colored pointmaps with language contrast produces unified 3D scene representations.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"b2c4a294ea30600cb92f0dee864ea52351c17d05a97586ced93c33ae320beede"},"source":{"id":"2604.02546","kind":"arxiv","version":2},"verdict":{"id":"7d850af7-33cf-4ef6-9e07-e700269ffec5","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T21:16:08.511554Z","strongest_claim":"We propose UniScene3D, a transformer-based encoder that learns unified scene representations from multi-view colored pointmaps, jointly modeling image appearance and geometry. ... Extensive low-shot and task-specific fine-tuning evaluations on viewpoint grounding, scene retrieval, scene type classification, and 3D VQA demonstrate our state-of-the-art performance.","one_line_summary":"UniScene3D learns unified 3D scene representations from colored pointmaps using contrastive CLIP pretraining plus cross-view geometric and grounded view alignments, achieving state-of-the-art results on viewpoint grounding, scene retrieval, classification, and 3D VQA.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the introduced cross-view geometric alignment and grounded view alignment will successfully enforce cross-view geometry and semantic consistency, leading to more generalizable unified representations (assumed without detailed verification of failure modes or data requirements in the abstract).","pith_extraction_headline":"Pretraining a transformer on multi-view colored pointmaps with language contrast produces unified 3D scene representations."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.02546/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","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"}