{"paper":{"title":"GemDepth: Geometry-Embedded Features for 3D-Consistent Video Depth","license":"http://creativecommons.org/licenses/by/4.0/","headline":"GemDepth achieves 3D-consistent video depth by predicting camera poses to embed geometric structure into a transformer.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hanrui Cheng, Junda Cheng, Longliang Liu, Wenjing Liao, Xin Yang, Yuecheng Liu, Yuzhou Wang","submitted_at":"2026-05-11T13:11:54Z","abstract_excerpt":"Video depth estimation extends monocular prediction into the temporal domain to ensure coherence. However, existing methods often suffer from spatial blurring in fine-detail regions and temporal inconsistencies. We argue that current approaches, which primarily rely on temporal smoothing via Transformers, struggle to maintain strict 3D geometric consistency-particularly under rotations or drastic view changes. To address this, we propose GemDepth, a framework built on the insight that an explicit awareness of camera motion and global 3D structure is a prerequisite for 3D consistency. Distincti"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"GemDepth achieves state-of-the-art performance across multiple datasets, particularly in complex dynamic scenarios.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That predicting inter-frame camera poses and injecting the resulting geometric embeddings is a prerequisite for strict 3D consistency under rotations or drastic view changes.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"GemDepth predicts inter-frame camera poses to inject geometric embeddings into a spatio-temporal transformer, yielding state-of-the-art 3D-consistent video depth.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"GemDepth achieves 3D-consistent video depth by predicting camera poses to embed geometric structure into a transformer.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"80438595b31431519e9ea6dba5fdfcec63c4fc113798ed2fe9aa4c191b71c601"},"source":{"id":"2605.10525","kind":"arxiv","version":4},"verdict":{"id":"67b9e8b1-0d64-4492-8b1e-b1ea42ed182c","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T21:24:42.864887Z","strongest_claim":"GemDepth achieves state-of-the-art performance across multiple datasets, particularly in complex dynamic scenarios.","one_line_summary":"GemDepth predicts inter-frame camera poses to inject geometric embeddings into a spatio-temporal transformer, yielding state-of-the-art 3D-consistent video depth.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That predicting inter-frame camera poses and injecting the resulting geometric embeddings is a prerequisite for strict 3D consistency under rotations or drastic view changes.","pith_extraction_headline":"GemDepth achieves 3D-consistent video depth by predicting camera poses to embed geometric structure into a transformer."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.10525/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-19T14:42:07.908786Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T11:01:17.869831Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T09:12:29.466117Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"e541b2a2e5568f4f98af6b2ca3dc7af5ac45400cd5819c7fbb84046f6aacb765"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"d92fc18f344a799a4924c04fa3abe32d116481688887298c9abf89a54530d1bb"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}