{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2607.09068", "kind": "paper", "url": "https://arxiv.org/abs/2607.09068"}, {"id": "gh:sigmme/omnimapbench", "kind": "repo", "url": "https://github.com/SIGMME/OmniMapBench"}], "authors": ["Yang Chen", "Yunwen Li", "Yufan Shen", "Minghao Liu", "Tianyu Zheng", "Bin Fu", "Qunshu Lin", "Zhi Yu", "Botian Shi"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.09068", "source_url": "https://arxiv.org/abs/2607.09068"}, "categories": ["General AI", "Reasoning", "Vision & 3D"], "description": {"en": "A visual question-answering benchmark of 2,096 manually annotated questions over 1,603 map documents in nine categories. It is scored by accuracy."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.09068", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.09068", "title": "OmniMapBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2607.09068", "modality": null, "name": "OmniMapBench", "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_omnimapbench_277894d3", "origin_source": "arxiv", "recognition_confidence": "0.75", "record_sha256": "d7feb0d1bbf80fa77e16a519db6cbb64bf72b6f9fc9b8c3226f536fc2c3ff5ed", "relation": "introduces", "review_model": "claude-code-default", "review_state": "ai-reviewed", "reviewed_at": "2026-09-24T16:34:01.178138Z", "source_url": "https://arxiv.org/abs/2607.09068"}, "publisher": null, "released": "2026-07-10", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.09068", "source_url": "https://arxiv.org/abs/2607.09068"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-09068", "source": "claire_radar", "source_benchmark_id": "2607.09068", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Vision & 3D", "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "A public LVLM benchmark with a 25-model evaluation and a clear accuracy metric. Its appeal is mainly to researchers working on document understanding.", "score": 45}, "capabilities": ["Reasoning"], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-24T16:34:01.178138Z", "inputHash": "68e664877b5e9982e48dc6ec68fc11d21aeaff1b02333a18605719b8162eaca6", "model": "claude-code-default"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "OmniMapBench: Benchmarking Visual-Centric Reasoning on Diverse Map Documents", "canonicalNameSource": "paper_title", "decisionReason": "A formally named benchmark with a fixed QA set, accuracy scoring, and a publicly released dataset and code.", "evaluationEvidence": {"reuse": "The dataset and code are publicly available at https://github.com/SIGMME/OmniMapBench.", "scoring": "the top-performing model achieving only 75.03\\% accuracy", "task": "2,096 manually annotated question-answer pairs across 1,603 map documents from nine categories"}, "model": "claude-code-default", "reviewedAt": "2026-09-24T16:34:01.178138Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A visual question-answering benchmark of 2,096 manually annotated questions over 1,603 map documents in nine categories. It is scored by accuracy.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "evaluation protocol evidence", "public artifact URL"], "snippet": "The benchmark comprises 2,096 manually annotated question-answer pairs across 1,603 map documents from nine categories."}, "familyId": "bmf_5e6fe95789f4", "firstSeenAt": "2026-08-19", "heat": null, "id": "bm_omnimapbench_277894d3", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/SIGMME/OmniMapBench", "data": null, "paper": "https://arxiv.org/abs/2607.09068", "pdf": "https://arxiv.org/pdf/2607.09068", "project": null, "report": "https://arxiv.org/abs/2607.09068"}, "metrics": [], "motivation": "Recent advancements in LVLMs necessitate robust benchmarks for complex, visually grounded reasoning.", "name": "OmniMapBench", "oneLine": "A visual question-answering benchmark of 2,096 manually annotated questions over 1,603 map documents in nine categories. It is scored by accuracy.", "paperTitle": "OmniMapBench: Benchmarking Visual-Centric Reasoning on Diverse Map Documents", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2607.09068", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-19T11:05:13.395391Z"}, "publications": [], "readiness": "Runnable", "recognitionConfidence": 0.75, "relation": "introduces", "releasedAt": "2026-07-10", "reviewLifecycle": {"state": "pending"}, "source": {"authors": ["Yang Chen and Yunwen Li and Yufan Shen and Minghao Liu and Tianyu Zheng and Bin Fu and Qunshu Lin and Zhi Yu and Botian Shi"], "categories": ["cs.CV", "cs.AI"], "id": "2607.09068", "title": "OmniMapBench: Benchmarking Visual-Centric Reasoning on Diverse Map Documents", "type": "arxiv", "url": "https://arxiv.org/abs/2607.09068"}, "sourceUpdatedAt": "2026-07-10T03:27:25Z", "topics": ["Reasoning"], "venueAttempts": [], "whyItMatters": "It targets map reasoning that cannot be reduced to text and adds a Visual Dependency Index to check that. The top model among 25 evaluated LVLMs reaches 75.03%."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
