{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2607.22393", "kind": "paper", "url": "https://arxiv.org/abs/2607.22393"}, {"id": "gh:feinaldo2/sceneactbench", "kind": "repo", "url": "https://github.com/Feinaldo2/SceneActBench"}], "authors": ["Yifei Zhao", "Xiangxin Zhou", "Wenhao Yang", "Jiaqi Tang", "Pu Jian", "Huanjin Yao", "Jiarui Yao", "Haowei Lin", "Chunchao Guo", "Zhuo Chen", "Wenkai Lyu", "Jianzhu Ma", "Xueqian Wang", "Wenxi Zhu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.22393", "source_url": "https://arxiv.org/abs/2607.22393"}, "categories": ["Agents", "Agents & Tool Use", "General AI", "Geometric reasoning"], "description": {"en": "A benchmark for visually conditioned action on complete multi-object 3D scenes across five tasks under a unified agent-environment loop, scored against hidden geometric ground truth."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.22393", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.22393", "title": "SceneActBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["University of Illinois Urbana-Champaign"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2607.22393", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2607.22393", "source_url": "https://openalex.org/W7171363301"}, "key": "claire-radar:2607.22393", "modality": null, "name": "SceneActBench", "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_sceneactbench_da42bbe6", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "d6ed5b206a87c796c730406cd1c3f742a8ecc2b01f505df2ccbaf165c09557d2", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2607.22393"}, "publisher": null, "released": "2026-07-24", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.22393", "source_url": "https://arxiv.org/abs/2607.22393"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-22393", "source": "claire_radar", "source_benchmark_id": "2607.22393", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Agents & Tool Use", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/Feinaldo2/SceneActBench", "githubScope": "benchmark_repo", "githubStars": 15, "hfDailySubmittedAt": "2026-07-27T00:00:00.000Z", "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 7, "hfPaperUrl": "https://huggingface.co/papers/2607.22393", "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": ["Geometric reasoning"], "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": "71c1287911aba64f84386cf017f29eb6f0743eb3d03f06eeaed0bc8953728f45", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A benchmark for visually conditioned action on complete multi-object 3D scenes across five tasks under a unified agent-environment loop, scored against hidden geometric ground truth.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We present SceneActBench, a benchmark for visually conditioned action across five 3D tasks under a unified agent-environment loop."}, "familyId": "bmf_9aedebf94723", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [{"evidence": [{"basis": "existing-code-link", "linkedUrl": "https://github.com/Feinaldo2/SceneActBench", "url": "https://github.com/Feinaldo2/SceneActBench"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/Feinaldo2/SceneActBench", "url": "https://huggingface.co/papers/2607.22393"}], "scope": "benchmark_repo", "url": "https://github.com/Feinaldo2/SceneActBench", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_sceneactbench_da42bbe6", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/Feinaldo2/SceneActBench", "data": null, "hfPaper": "https://huggingface.co/papers/2607.22393", "paper": "https://arxiv.org/abs/2607.22393", "pdf": "https://arxiv.org/pdf/2607.22393", "project": null, "report": "https://arxiv.org/abs/2607.22393"}, "metrics": [], "motivation": "Vision-language model (VLM) agents increasingly use tools to act on 3D scenes rather than only describe them.", "name": "SceneActBench", "oneLine": "A benchmark for visually conditioned action on complete multi-object 3D scenes across five tasks under a unified agent-environment loop, scored against hidden geometric ground truth.", "paperTitle": "SceneActBench: Can Agents Act on the 3D Scenes They See?", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2607.22393", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "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": 0.3700907540283743, "value": 15}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.3674261111536563, "value": 7}}, "confidence": "Medium", "coverage": 0.7, "rank": 133, "score": 41}, "method": "attention-ranking-v13", "rank": 133, "score": 41}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-07-24", "source": {"authors": ["Yifei Zhao", "Xiangxin Zhou", "Wenhao Yang", "Jiaqi Tang", "Pu Jian", "Huanjin Yao", "Jiarui Yao", "Haowei Lin", "Chunchao Guo", "Zhuo Chen", "Wenkai Lyu", "Jianzhu Ma", "Xueqian Wang", "Wenxi Zhu"], "categories": ["cs.AI", "cs.CV"], "id": "2607.22393", "title": "SceneActBench: Can Agents Act on the 3D Scenes They See?", "type": "arxiv", "url": "https://arxiv.org/abs/2607.22393"}, "sourceUpdatedAt": "2026-07-24T15:16:47Z", "topics": ["Agents"], "venueAttempts": [], "whyItMatters": "Addresses the under-evaluation of agent action on complete 3D scenes, providing a comparative basis for VLM agents acting on 3D environments."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
