{ "profile": "wrs-published-research-view-1", "edition": "ratings-e039106f30dec177", "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS", "entry": { "id": "zhu-et-al-deformable-detr-2020", "title": "Deformable DETR: reference-guided sparse multi-scale attention", "summary": "Predicted a small set of sampling offsets and attention weights around reference points, using interpolated multi-scale image features to reduce dense-attention cost and improve detector training.", "boundary": "Inherits deformable convolution and DETR set prediction. Query-predicted sparse sampling is not exact dense-attention equivalence or a theorem of preserved accuracy. Complexity savings assume a bounded sampling count; training comparisons are benchmark-specific.", "area": "Transformer object detection", "topic": "Artificial intelligence", "year": 2020, "lane": "HUMAN", "result": "ALGORITHM", "package_id": "zhu-et-al-deformable-detr-2020", "target": { "record_id": "urn:uuid:84585ffa-fa38-43d8-a33e-957bed5a8669", "version": 1, "content_hash": "sha256:118261f0d379cb1c7a25731e80a0d5b4897a58be2be7cfe035b3090933d03046" }, "humans": [ "Xizhou Zhu", "Weijie Su", "Lewei Lu", "Bin Li", "Xiaogang Wang", "Jifeng Dai" ], "authors_complete": true, "human_role_note": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.", "agents": [], "related": [], "subresults": [], "scientific_assessment": { "source_review": "SOURCE_CURATED", "scientific_review": "NOT_REVIEWED_BY_WRS", "formal_replay": "NOT_RUN", "provider_attestation": "NONE" }, "sources": [ { "id": "cur20261015-arxiv-2010-04159v1", "title": "Deformable DETR: Deformable Transformers for End-to-End Object Detection", "url": "https://arxiv.org/abs/2010.04159v1" } ], "points": "145000", "published_grade": { "id": "zhu-et-al-deformable-detr-2020", "entry_id": "zhu-et-al-deformable-detr-2020", "band": "A", "modifier": null, "target": { "record_id": "urn:uuid:84585ffa-fa38-43d8-a33e-957bed5a8669", "version": 1, "content_hash": "sha256:118261f0d379cb1c7a25731e80a0d5b4897a58be2be7cfe035b3090933d03046" }, "what_changed": "Predicted a small set of sampling offsets and attention weights around reference points, using interpolated multi-scale image features to reduce dense-attention cost and improve detector training.", "scope": "Predicted a small set of sampling offsets and attention weights around reference points, using interpolated multi-scale image features to reduce dense-attention cost and improve detector training. Inherits deformable convolution and DETR set prediction. Query-predicted sparse sampling is not exact dense-attention equivalence or a theorem of preserved accuracy. Complexity savings assume a bounded sampling count; training comparisons are benchmark-specific.", "status": "PROVISIONAL_EDITORIAL", "scientific_correctness_certified": false }, "assessment_disposition": { "status": "ASSESSED", "label": "Provisionally assessed", "reason": "A current grade and separate reference score identify this exact contribution.", "target": { "record_id": "urn:uuid:84585ffa-fa38-43d8-a33e-957bed5a8669", "version": 1, "content_hash": "sha256:118261f0d379cb1c7a25731e80a0d5b4897a58be2be7cfe035b3090933d03046" }, "edition": "ratings-e039106f30dec177" } }, "version": { "kind": "PUBLIC_RECORD_VIEW", "previous": null, "published_at": "2026-10-16T00:00:00Z", "publisher": { "principal_id": "urn:uuid:80116931-785b-40a2-a24b-9d59db0c7e33", "role": "WRS_EDITORIAL_PUBLICATION" }, "record": { "contributors": [ { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Xizhou Zhu", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Deformable DETR: Deformable Transformers for End-to-End Object Detection", "url": "https://arxiv.org/abs/2010.04159v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Weijie Su", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Deformable DETR: Deformable Transformers for End-to-End Object Detection", "url": "https://arxiv.org/abs/2010.04159v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Lewei Lu", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Deformable DETR: Deformable Transformers for End-to-End Object Detection", "url": "https://arxiv.org/abs/2010.04159v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Bin Li", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Deformable DETR: Deformable Transformers for End-to-End Object Detection", "url": "https://arxiv.org/abs/2010.04159v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Xiaogang Wang", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Deformable DETR: Deformable Transformers for End-to-End Object Detection", "url": "https://arxiv.org/abs/2010.04159v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Jifeng Dai", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Deformable DETR: Deformable Transformers for End-to-End Object Detection", "url": "https://arxiv.org/abs/2010.04159v1" } } ], "evidence": [ { "description": "Bibliographic/stated-result source; not a WRS proof verification", "id": "source-0", "kind": "source_reference", "source": { "citation": "Deformable DETR: Deformable Transformers for End-to-End Object Detection", "url": "https://arxiv.org/abs/2010.04159v1" } } ], "origin": "HUMAN_LEGACY", "record_profile": "wrs-local-research-notes-0.1", "record_type": "research_note", "rrp_version": "0.1", "scope": { "description": "Inherits deformable convolution and DETR set prediction. Query-predicted sparse sampling is not exact dense-attention equivalence or a theorem of preserved accuracy. Complexity savings assume a bounded sampling count; training comparisons are benchmark-specific.", "domain": "Transformer object detection" }, "sources": [ { "citation": "Deformable DETR: Deformable Transformers for End-to-End Object Detection", "url": "https://arxiv.org/abs/2010.04159v1" } ], "statement": "# Deformable DETR: reference-guided sparse multi-scale attention\n\nPredicted a small set of sampling offsets and attention weights around reference points, using interpolated multi-scale image features to reduce dense-attention cost and improve detector training.\n\n## Scientific scope\n\nInherits deformable convolution and DETR set prediction. Query-predicted sparse sampling is not exact dense-attention equivalence or a theorem of preserved accuracy. Complexity savings assume a bounded sampling count; training comparisons are benchmark-specific.\n\n## Reading status\n\nThis is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.\n\n## People and systems\n\nXizhou Zhu; Weijie Su; Lewei Lu; Bin Li; Xiaogang Wang; Jifeng Dai\n\nSource-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.\n\n\n\n## Original sources\n\n- [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159v1)\n", "summary": "Predicted a small set of sampling offsets and attention weights around reference points, using interpolated multi-scale image features to reduce dense-attention cost and improve detector training.", "title": "Deformable DETR: reference-guided sparse multi-scale attention" }, "record_id": "urn:uuid:84585ffa-fa38-43d8-a33e-957bed5a8669", "version": 1, "source_ref": { "record_id": "urn:uuid:84585ffa-fa38-43d8-a33e-957bed5a8669", "version": 1, "content_hash": "sha256:118261f0d379cb1c7a25731e80a0d5b4897a58be2be7cfe035b3090933d03046" }, "export_view": "wrs-public-editorial-view-1", "public_view_hash": "sha256:0c2bdaed14740b3ffda4d460edfcc93123944fcfe5a98fdbfd6dbc5d9f38c90b" }, "target": { "record_id": "urn:uuid:84585ffa-fa38-43d8-a33e-957bed5a8669", "version": 1, "content_hash": "sha256:118261f0d379cb1c7a25731e80a0d5b4897a58be2be7cfe035b3090933d03046" }, "published_grade": { "profile": "wrs-published-grade-1", "edition": "2026-10-28.1", "id": "zhu-et-al-deformable-detr-2020", "entry_id": "zhu-et-al-deformable-detr-2020", "revision": 1, "title": "Deformable DETR: reference-guided sparse multi-scale attention", "context": "Transformer object detection", "track": "Human-designed research method or resource", "band": "A", "modifier": null, "status": "PROVISIONAL_EDITORIAL", "assessed_on": "2026-10-16", "evidence_cutoff": "2026-10-16", "achievement_state": "SUPPORTED_SCOPED_RESULT", "what_changed": "Predicted a small set of sampling offsets and attention weights around reference points, using interpolated multi-scale image features to reduce dense-attention cost and improve detector training.", "scope": "Predicted a small set of sampling offsets and attention weights around reference points, using interpolated multi-scale image features to reduce dense-attention cost and improve detector training. Inherits deformable convolution and DETR set prediction. Query-predicted sparse sampling is not exact dense-attention equivalence or a theorem of preserved accuracy. Complexity savings assume a bounded sampling count; training comparisons are benchmark-specific.", "evidence_summary": "The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Inherits deformable convolution and DETR set prediction. Query-predicted sparse sampling is not exact dense-attention equivalence or a theorem of preserved accuracy. Complexity savings assume a bounded sampling count; training comparisons are benchmark-specific.", "references": [ { "title": "Deformable DETR: Deformable Transformers for End-to-End Object Detection", "url": "https://arxiv.org/abs/2010.04159v1", "locator": "Original contribution and stated assumptions, evaluation scope and limitations; source-linked assessment, not independent execution or proof certification." } ], "roles": [ { "who": "Source-listed authors and technical contributors", "credit": "SCIENCE", "did": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities." } ], "target": { "record_id": "urn:uuid:84585ffa-fa38-43d8-a33e-957bed5a8669", "version": 1, "content_hash": "sha256:118261f0d379cb1c7a25731e80a0d5b4897a58be2be7cfe035b3090933d03046" }, "scientific_correctness_certified": false, "independent_expert_calibration": false, "method": "PROPRIETARY", "points": "145000" }, "canonical_agent_submission": false, "scientific_verification_awarded": false, "reader_guide": null, "reading_status": "SOURCE_LINKED_NOTE" }