{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2607.06420", "kind": "paper", "url": "https://arxiv.org/abs/2607.06420"}, {"id": "gh:mm-mvr/holocount", "kind": "repo", "url": "https://github.com/MM-MVR/HoloCount"}, {"kind": "website", "url": "https://mm-mvr.github.io/HoloCount/"}], "authors": ["Jinhong Deng", "Limeng Qiao", "Guanglu Wan"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.06420", "source_url": "https://arxiv.org/abs/2607.06420"}, "categories": ["General AI", "Vision & 3D", "cs.CV"], "description": {"en": "HoloCount evaluates multimodal large language models on visual counting tasks across three levels: semantic counting (atomic and property-based enumeration), analytical counting (logical composition via spatial and set-based reasoning), and robustness testing (adverse scenarios and grounded counter-priors). The benchmark uses a hierarchical taxonomy and provides a dataset for evaluation."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.06420", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.06420", "title": "HoloCount"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2607.06420", "modality": null, "name": "HoloCount", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "unknown", "origin_record_id": "bm_holocount_b6a34c73", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "7e9bdca0bd2e324f1f1eefdacd6f07b7e4432c7c7e64b837a0c6b8872944e83b", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2607.06420"}, "publisher": null, "released": "2026-07-07", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.06420", "source_url": "https://arxiv.org/abs/2607.06420"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-06420", "source": "claire_radar", "source_benchmark_id": "2607.06420", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Vision & 3D", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/MM-MVR/HoloCount", "githubScope": "benchmark_repo", "githubStars": 2, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 0, "hfPaperUrl": "https://huggingface.co/papers/2607.06420", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}}, "status": "fresh"}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "8a5aecf80c7de9ed617c8b644b55b7fd52e8c32f92a01db2e358afaad9661c72", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "HoloCount evaluates multimodal large language models on visual counting tasks across three levels: semantic counting (atomic and property-based enumeration), analytical counting (logical composition via spatial and set-based reasoning), and robustness testing (adverse scenarios and grounded counter-priors). The benchmark uses a hierarchical taxonomy and provides a dataset for evaluation.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "To address these limitations, we introduce HoloCount, a holistic and diagnostically rich benchmark structured around a three-level hierarchical taxonomy."}, "familyId": "bmf_bd0b3a3f9aa2", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 1, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [{"evidence": [{"basis": "linked-from-project", "linkedUrl": "https://github.com/MM-MVR/HoloCount", "url": "https://mm-mvr.github.io/HoloCount/"}], "scope": "benchmark_repo", "url": "https://github.com/MM-MVR/HoloCount", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_holocount_b6a34c73", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/MM-MVR/HoloCount", "data": null, "hfPaper": "https://huggingface.co/papers/2607.06420", "paper": "https://arxiv.org/abs/2607.06420", "pdf": "https://arxiv.org/pdf/2607.06420", "project": "https://mm-mvr.github.io/HoloCount/", "report": "https://arxiv.org/abs/2607.06420"}, "metrics": [], "motivation": "Visual counting is a fundamental pillar of multimodal intelligence, requiring a seamless integration of fine-grained grounding and spatial reasoning.", "name": "HoloCount", "oneLine": "HoloCount evaluates multimodal large language models on visual counting tasks across three levels: semantic counting (atomic and property-based enumeration), analytical counting (logical composition via spatial and set-based reasoning), and robustness testing (adverse scenarios and grounded counter-priors). The benchmark uses a hierarchical taxonomy and provides a dataset for evaluation.", "paperTitle": "HoloCount: A Holistic Visual Counting Benchmark for MLLMs", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2607.06420", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "HoloCount Team", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://mm-mvr.github.io/HoloCount/"}], "ranking": {"90d": {"confidence": "Medium", "coverage": 0.7, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": 0.1466449917496476, "value": 2}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.0, "value": 0}}, "confidence": "Medium", "coverage": 0.7, "rank": 382, "score": 23}, "method": "attention-ranking-v13", "rank": 382, "score": 23}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-07-07", "source": {"authors": ["Jinhong Deng", "Limeng Qiao", "Guanglu Wan"], "categories": ["cs.CV", "cs.AI"], "id": "2607.06420", "title": "HoloCount: A Holistic Visual Counting Benchmark for MLLMs", "type": "arxiv", "url": "https://arxiv.org/abs/2607.06420"}, "sourceUpdatedAt": "2026-07-07T15:48:57Z", "topics": ["cs.CV"], "venueAttempts": [], "whyItMatters": "Existing counting benchmarks fail to capture complex failure modes under logical constraints or adversarial conditions. HoloCount provides a structured diagnostic tool to assess MLLM quantitative precision, revealing performance gaps as tasks shift from perception to analytical reasoning, and guiding development of more grounded multimodal systems."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
