{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.25491", "kind": "paper", "url": "https://arxiv.org/abs/2606.25491"}], "authors": ["Chuangxin Zhao", "Boyan Shi", "Yanling Wang", "Yijian LU", "Canran Xiao", "Jiali Chen", "Jun Xia", "Yan Wang", "Ji Qi", "Juanzi Li"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.25491", "source_url": "https://arxiv.org/abs/2606.25491"}, "categories": ["General AI", "Vision & 3D", "cs.CV"], "description": {"en": "HG-Bench evaluates page-aware, two-level answer-region grounding: given multi-page handwritten homework images, models must localize complete answer regions and ordered step-level subregions, with question- and step-level boxes under a hierarchical constraint."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.25491", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.25491", "title": "HG-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.25491", "modality": null, "name": "HG-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_hg-bench_1e72857e", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "f6f97c45f1193f9ba87a03d900d91123d00f48f29a81cfd9c58d35ebabfe2b0b", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.25491"}, "publisher": null, "released": "2026-06-24", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.25491", "source_url": "https://arxiv.org/abs/2606.25491"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-25491", "source": "claire_radar", "source_benchmark_id": "2606.25491", "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": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2606.25491", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "unavailable"}, "capabilities": [], "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": "39c9fa1786310ab51bdd5cc4043c56cafc1d7cb28dc7a2124c791f16256ecc89", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "HG-Bench evaluates page-aware, two-level answer-region grounding: given multi-page handwritten homework images, models must localize complete answer regions and ordered step-level subregions, with question- and step-level boxes under a hierarchical constraint.", "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 introduce HG-Bench, a benchmark of 500 human-annotated K-12 homework samples curated from a 1,489,278-image source pool, with question-level and step-level boxes linked by a hierarchical containment constraint."}, "familyId": "bmf_5e2cc69b8f13", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_hg-bench_1e72857e", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.25491", "paper": "https://arxiv.org/abs/2606.25491", "pdf": "https://arxiv.org/pdf/2606.25491", "project": null, "report": "https://arxiv.org/abs/2606.25491"}, "metrics": [], "motivation": "Automated homework assessment depends not only on recognizing student answers, but also on accurately locating where each answer and each intermediate reasoning step appears in noisy, multi-page handwritten work.", "name": "HG-Bench", "oneLine": "HG-Bench evaluates page-aware, two-level answer-region grounding: given multi-page handwritten homework images, models must localize complete answer regions and ordered step-level subregions, with question- and step-level boxes under a hierarchical constraint.", "paperTitle": "HG-Bench: A Benchmark for Multi-Page Handwritten Answer-Region Grounding in Automated Homework Assessment", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.25491", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "arXiv", "organizationType": "community", "role": "benchmark-publisher", "sourceUrl": "https://arxiv.org/abs/2606.25491"}], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-24", "source": {"authors": ["Chuangxin Zhao", "Boyan Shi", "Yanling Wang", "Yijian LU", "Canran Xiao", "Jiali Chen", "Jun Xia", "Yan Wang", "Ji Qi", "Juanzi Li"], "categories": ["cs.CV", "cs.AI"], "id": "2606.25491", "title": "HG-Bench: A Benchmark for Multi-Page Handwritten Answer-Region Grounding in Automated Homework Assessment", "type": "arxiv", "url": "https://arxiv.org/abs/2606.25491"}, "sourceUpdatedAt": "2026-06-24T07:18:50Z", "topics": ["cs.CV"], "venueAttempts": [], "whyItMatters": "Automated homework assessment needs both answer recognition and spatial grounding of reasoning steps. Prior benchmarks miss page-aware, multi-level localization; HG-Bench provides a reproducible protocol to measure this capability gap across models."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
