{"record": {"aliases": ["SpatialQuery: Benchmarking Geometry-Grounded Multi-Instance Spatial Reasoning in Vision-Language Models"], "artifacts": [{"id": "arxiv:2608.01709", "kind": "paper", "url": "https://arxiv.org/abs/2608.01709"}, {"kind": "website", "url": "https://namhai1810.github.io/SpatialQuery/"}], "categories": ["General AI", "Geometric reasoning", "Multimodal", "Reasoning"], "description": {"en": "A benchmark of over one million RGB-only question-answer pairs from 200 indoor scenes for closest-instance distance queries in vision-language models."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.01709", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.01709", "title": "SPATIALQUERY-1M"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.01709", "modality": null, "name": "SPATIALQUERY-1M", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-09-25T04:40:22Z", "data_status": "primary-source-indexed", "display_eligible": "true", "origin_record_id": "bm_spatialquery-benchmarking-geometry-grounde_078e6435", "origin_source": "arxiv", "recognition_confidence": "0.55", "record_sha256": "589368adddd9d8bf3221f9848f8beff215211d83082c3edd615545faa1c66878", "relation": "evaluates_only", "review_model": "claude-code-default", "review_state": "ai-reviewed", "reviewed_at": "2026-09-24T20:55:04.887078Z", "source_url": "https://arxiv.org/abs/2608.01709"}, "publisher": null, "released": "2026-08-03", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.01709", "source_url": "https://arxiv.org/abs/2608.01709"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-01709", "source": "claire_radar", "source_benchmark_id": "2608.01709", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": ["SpatialQuery: Benchmarking Geometry-Grounded Multi-Instance Spatial Reasoning in Vision-Language Models"], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Multimodal", "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "Spatial reasoning in VLMs gets steady interest; the large benchmark and public resources may help it get used.", "score": 40}, "capabilities": ["Reasoning", "Geometric reasoning"], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-24T20:55:04.887078Z", "inputHash": "0ccf6229be69ea5f334f6796b5a8b21f33e8126afdb338f52e1097bde8e1de67", "model": "claude-code-default"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "SPATIALQUERY-1M, a benchmark containing over one million RGB-only question-answer pairs from 200 indoor scenes", "canonicalNameSource": "abstract", "decisionReason": "Named benchmark with a defined CIDQ task, concrete metrics (Floor-MAE, Unc-Acc@0.3 m, proximity-decision accuracy), and a public project link to benchmark resources.", "evaluationEvidence": {"reuse": "Code, benchmark resources, and an interactive demo are available at https://namhai1810.github.io/SpatialQuery/", "scoring": "a Floor-MAE of 0.259 m, an Unc-Acc@0.3 m of 90.5%, and a proximity-decision accuracy of 84.18%", "task": "a model must identify the nearest visible candidate to a unique reference object and estimate their gravity-aligned floor-plane distance"}, "model": "claude-code-default", "reviewedAt": "2026-09-24T20:55:04.887078Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A benchmark of over one million RGB-only question-answer pairs from 200 indoor scenes for closest-instance distance queries in vision-language models.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["benchmark term in abstract", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL", "named entity is a model, method, framework, or dataset used in evaluation"], "snippet": "We introduce SPATIALQUERY, a training- free framework for CIDQ reasoning from a single RGB image, together with SPATIALQUERY-1M, a benchmark containing over one million RGB-only question-answer pairs from 200 indoor scenes."}, "familyId": "bmf_95929b60b5d9", "firstSeenAt": "2026-08-19", "heat": null, "id": "bm_spatialquery-benchmarking-geometry-grounde_078e6435", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "paper": "https://arxiv.org/abs/2608.01709", "pdf": "https://arxiv.org/pdf/2608.01709", "project": "https://namhai1810.github.io/SpatialQuery/", "report": "https://arxiv.org/abs/2608.01709"}, "metrics": [], "motivation": "Vision-language models (VLMs) achieve strong semantic understanding but remain unreliable in metric spatial reasoning, particularly when queries require comparing multiple instances of the same object category.", "name": "SPATIALQUERY-1M", "oneLine": "A benchmark of over one million RGB-only question-answer pairs from 200 indoor scenes for closest-instance distance queries in vision-language models.", "paperTitle": "SpatialQuery: Benchmarking Geometry-Grounded Multi-Instance Spatial Reasoning in Vision-Language Models", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.01709", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-19T11:05:13.395391Z"}, "publications": [], "readiness": "Paper only", "recognitionConfidence": 0.55, "relation": "evaluates_only", "releasedAt": "2026-08-03", "reviewLifecycle": {"state": "pending"}, "source": {"authors": ["Hai Nguyen", "Tung Vu", "Cong Tran"], "categories": ["cs.CV"], "id": "2608.01709", "title": "SpatialQuery: Benchmarking Geometry-Grounded Multi-Instance Spatial Reasoning in Vision-Language Models", "type": "arxiv", "url": "https://arxiv.org/abs/2608.01709"}, "sourceUpdatedAt": "2026-08-03T05:15:46Z", "topics": ["Multimodal", "Reasoning"], "venueAttempts": [], "whyItMatters": "It targets multi-instance metric spatial reasoning, where VLMs remain unreliable, and scores models with concrete distance-error and decision-accuracy metrics."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
