{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2607.02680", "kind": "paper", "url": "https://arxiv.org/abs/2607.02680"}, {"id": "gh:ogmenrobotics/k9bench-eval-code", "kind": "repo", "url": "https://github.com/ogmenrobotics/K9Bench-eval-code"}, {"kind": "website", "url": "https://ogmenrobotics.github.io/K9Bench"}], "authors": ["Khush Attarde", "Yusuf Ali", "Megha Thukral", "Divye Bhutani", "Thomas Ploetz", "Zsolt Kira"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.02680", "source_url": "https://arxiv.org/abs/2607.02680"}, "categories": ["Multimodal", "Reasoning", "Robotics & Embodied AI"], "description": {"en": "K9-Bench is a benchmark focused on real-world domestic dog videos, with approximately 5,000 question-answer pairs across 907 videos spanning 5 task categories. The tasks test long-form, canine-centric multimodal reasoning, including action and interaction understanding. A VLM/LLM-powered pipeline was used for data generation."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.02680", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.02680", "title": "K9-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2607.02680", "modality": null, "name": "K9-Bench", "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_k9-bench_cbdfe117", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "57bb305d3d5fe04fdac5e0bd51babdcf76dbbb891d1a82a5c8fde872b5fea2d5", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2607.02680"}, "publisher": null, "released": "2026-07-02", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.02680", "source_url": "https://arxiv.org/abs/2607.02680"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-02680", "source": "claire_radar", "source_benchmark_id": "2607.02680", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Robotics & Embodied AI"], "area": "Multimodal", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/ogmenrobotics/K9Bench-eval-code", "githubScope": "benchmark_repo", "githubStars": 1, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 0, "hfPaperUrl": "https://huggingface.co/papers/2607.02680", "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": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-10-04", "state": "stale"}}, "status": "fresh"}, "capabilities": ["Reasoning"], "confidence": "High", "construction": "Transform Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "c4078098e6a609a824fa7ca61f8db7dfa08f76da36f9547bbbda889098c28934", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "K9-Bench is a benchmark focused on real-world domestic dog videos, with approximately 5,000 question-answer pairs across 907 videos spanning 5 task categories. The tasks test long-form, canine-centric multimodal reasoning, including action and interaction understanding. A VLM/LLM-powered pipeline was used for data generation.", "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", "public artifact URL"], "snippet": "We introduce K9-Bench, a novel benchmark focused on real-world domestic dog videos, specifically targeting canine action and interaction understanding via approximately 5000 question-answer pairs across 907 videos spanning 5 distinct task categories that test long-form, canine-centric multimodal reasoning in MLLMs."}, "familyId": "bmf_b14e23a954dd", "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/ogmenrobotics/K9Bench-eval-code", "url": "https://ogmenrobotics.github.io/K9Bench"}], "scope": "benchmark_repo", "url": "https://github.com/ogmenrobotics/K9Bench-eval-code", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_k9-bench_cbdfe117", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Robotics"], "links": {"code": "https://github.com/ogmenrobotics/K9Bench-eval-code", "data": null, "hfPaper": "https://huggingface.co/papers/2607.02680", "paper": "https://arxiv.org/abs/2607.02680", "pdf": "https://arxiv.org/pdf/2607.02680", "project": "https://ogmenrobotics.github.io/K9Bench", "report": "https://arxiv.org/abs/2607.02680"}, "metrics": [], "motivation": "MLLMs have shown strong zero-shot capabilities across diverse inputs such as across images, video, audio, and text.", "name": "K9-Bench", "oneLine": "K9-Bench is a benchmark focused on real-world domestic dog videos, with approximately 5,000 question-answer pairs across 907 videos spanning 5 task categories. The tasks test long-form, canine-centric multimodal reasoning, including action and interaction understanding. A VLM/LLM-powered pipeline was used for data generation.", "paperTitle": "K9-Bench: Evaluating Multimodal LLMs on Canine-Centric Videos", "primaryDomain": "Robotics & Embodied AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2607.02680", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-07-02", "source": {"authors": ["Khush Attarde", "Yusuf Ali", "Megha Thukral", "Divye Bhutani", "Thomas Ploetz", "Zsolt Kira"], "categories": ["cs.CV", "cs.AI"], "id": "2607.02680", "title": "K9-Bench: Evaluating Multimodal LLMs on Canine-Centric Videos", "type": "arxiv", "url": "https://arxiv.org/abs/2607.02680"}, "sourceUpdatedAt": "2026-07-02T18:23:14Z", "topics": ["Multimodal", "Reasoning"], "venueAttempts": [], "whyItMatters": "It addresses the underexplored application of multimodal models to animal-centric scenarios, providing a benchmark to evaluate models on recognizing distress signals and enabling responsive robotic companions."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
