{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.29607", "kind": "paper", "url": "https://arxiv.org/abs/2608.29607"}], "authors": ["Zirong Chen", "Fuda Ye", "Kuan Zhang", "Enjun Du", "Junfu Pu", "Xinlei Wang", "Xinyu Zuo", "Lisheng Duan", "Jin Ma", "Yongqi Zhang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.29607", "source_url": "https://arxiv.org/abs/2608.29607"}, "categories": ["General AI", "Information retrieval", "Multimodal"], "description": {"en": "Evaluates multimodal retrieval on 1,145 queries and 9,085 gallery items under 53 controlled image and text corruption conditions, with clean and noisy matching metrics."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.29607", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.29607", "title": "SnapBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.29607", "modality": null, "name": "SnapBench", "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": "true", "origin_record_id": "bm_snapbench_c48382dd", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "5932e4b073061c882fb1dade0796016d2ba02598b7de448870c94639d2dd0b7d", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-09-02T05:23:11.715633Z", "source_url": "https://arxiv.org/abs/2608.29607"}, "publisher": null, "released": "2026-08-30", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.29607", "source_url": "https://arxiv.org/abs/2608.29607"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-29607", "source": "claire_radar", "source_benchmark_id": "2608.29607", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Multimodal", "attention": {"asOf": "2026-09-02", "attemptedAt": "2026-09-02T05:12:51.621571Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 1, "hfPaperUrl": "https://huggingface.co/papers/2608.29607", "observedAt": "2026-09-02T05:12:51.621571Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"observedAt": "2026-09-02T05:12:51.621571Z", "state": "fresh"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "Mobile multimodal retrieval with realistic corruptions is a broadly relevant topic, and the benchmark includes a novel paired corruption design and 16 evaluator models, likely to attract attention from both retrieval and embodied AI communities.", "score": 72}, "capabilities": ["Information retrieval"], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-02T05:23:11.715633Z", "inputHash": "48093e1ddf7ba199168a7472795278e6ef4f6d09e96e714933bf6d5486e1b0d6", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "Therefore, we introduce SnapBench, the first paired benchmark for robust snap-and-ask multimodal retrieval", "canonicalNameSource": "abstract", "decisionReason": "The paper introduces a named benchmark with a defined query/gallery/corruption protocol and evaluates 16 retrievers, providing a repeatable scoring setup and a citable paper as a public path.", "model": "deepseek-v4-pro", "reviewedAt": "2026-09-02T05:23:11.715633Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Evaluates multimodal retrieval on 1,145 queries and 9,085 gallery items under 53 controlled image and text corruption conditions, with clean and noisy matching metrics.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "Therefore, we introduce SnapBench, the first paired benchmark for robust snap-and-ask multimodal retrieval, spanning 1,145 queries, 9,085 gallery items under 53 controlled corruption conditions with human annotations."}, "familyId": "bmf_51cbb8fc6f60", "firstSeenAt": "2026-09-01", "heat": null, "id": "bm_snapbench_c48382dd", "indexedAt": "2026-09-02T05:21:06.631990Z", "industrySectors": [], "links": {"code": null, "data": null, "paper": "https://arxiv.org/abs/2608.29607", "pdf": "https://arxiv.org/pdf/2608.29607", "project": null, "report": "https://arxiv.org/abs/2608.29607"}, "metrics": [], "motivation": "Mobile AI acts as a visual oracle, empowering users to snap a picture of something and ask for information.", "name": "SnapBench", "oneLine": "Evaluates multimodal retrieval on 1,145 queries and 9,085 gallery items under 53 controlled image and text corruption conditions, with clean and noisy matching metrics.", "paperTitle": "SnapBench: Benchmarking Snap-and-Ask Multimodal Retrieval for Mobile Interactions", "primaryDomain": "General AI", "publication": {"evidence": "37 pages. Yuanbao Technical Report. Accepted to Findings of EMNLP 2026", "evidenceLevel": "author-claim", "evidenceUrl": "https://arxiv.org/abs/2608.29607", "source": "arxiv-comments", "status": "acceptance_claimed", "venue": "Findings of EMNLP 2026", "verifiedAt": "2026-09-02T05:21:06.631990Z"}, "publications": [], "publishers": [{"name": "SnapBench Team", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://arxiv.org/abs/2608.29607"}], "ranking": {"30d": {"confidence": "Low", "coverage": 0.3, "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": 30}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.1228568942814387, "value": 1}}, "confidence": "Low", "coverage": 0.3, "rank": null, "score": 39}, "method": "attention-ranking-v13", "rank": null, "score": 39}, "90d": {"confidence": "Low", "coverage": 0.15, "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": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.1228568942814387, "value": 1}}, "confidence": "Low", "coverage": 0.15, "rank": null, "score": 44}, "method": "attention-ranking-v13", "rank": null, "score": 44}}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-08-30", "source": {"authors": ["Zirong Chen", "Fuda Ye", "Kuan Zhang", "Enjun Du", "Junfu Pu", "Xinlei Wang", "Xinyu Zuo", "Lisheng Duan", "Jin Ma", "Yongqi Zhang"], "categories": ["cs.CV", "cs.IR"], "id": "2608.29607", "title": "SnapBench: Benchmarking Snap-and-Ask Multimodal Retrieval for Mobile Interactions", "type": "arxiv", "url": "https://arxiv.org/abs/2608.29607"}, "sourceUpdatedAt": "2026-08-30T06:56:11Z", "topics": ["Multimodal"], "venueAttempts": [{"decisionRaw": "37 pages. Yuanbao Technical Report. Accepted to Findings of EMNLP 2026", "evidence": [{"level": "author-claim", "observedAt": "2026-09-02T05:21:06.631990Z", "rawValue": "37 pages. Yuanbao Technical Report. Accepted to Findings of EMNLP 2026", "sourceType": "arxiv-comments", "sourceUrl": "https://arxiv.org/abs/2608.29607"}], "reviewStatus": "accepted", "venueName": "Findings of EMNLP 2026"}], "whyItMatters": "Snap-and-ask mobile retrieval faces noisy photos and short/typo text, yet existing benchmarks test only clean inputs; SnapBench isolates corruption effects and shows clean image-only retrieval often beats joint retrieval, informing robust system design."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
