{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:IDKVDK63DOKR37PADZNMOLEQ2L","short_pith_number":"pith:IDKVDK63","schema_version":"1.0","canonical_sha256":"40d551abdb1b951dfde01e5ac72c90d2c01ab2b229a10cbd6446af01942010ab","source":{"kind":"arxiv","id":"2205.08508","version":1},"attestation_state":"computed","paper":{"title":"A CLIP-Hitchhiker's Guide to Long Video Retrieval","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Andrew Zisserman, Arsha Nagrani, G\\\"ul Varol, Max Bain","submitted_at":"2022-05-17T17:26:23Z","abstract_excerpt":"Our goal in this paper is the adaptation of image-text models for long video retrieval. Recent works have demonstrated state-of-the-art performance in video retrieval by adopting CLIP, effectively hitchhiking on the image-text representation for video tasks. However, there has been limited success in learning temporal aggregation that outperform mean-pooling the image-level representations extracted per frame by CLIP. We find that the simple yet effective baseline of weighted-mean of frame embeddings via query-scoring is a significant improvement above all prior temporal modelling attempts and"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2205.08508","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2022-05-17T17:26:23Z","cross_cats_sorted":[],"title_canon_sha256":"b94981acca69d546b1fe772bb5ce6373d78c2c6ede8d0a1f8df848df4092ec0f","abstract_canon_sha256":"b82a53da9182e6249adf84474d6411d0d8158a01fca878cf68873fb9ab831863"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:24:03.800808Z","signature_b64":"HIKL7qHfy7jzTsLl+5a5A6MMCfAdyNxD9HlQy7j1JCbJ/7KZgLivL3noM/PnTZTQTeOgb2GRPIQZPKZU9Z5JDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"40d551abdb1b951dfde01e5ac72c90d2c01ab2b229a10cbd6446af01942010ab","last_reissued_at":"2026-07-05T04:24:03.800288Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:24:03.800288Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A CLIP-Hitchhiker's Guide to Long Video Retrieval","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Andrew Zisserman, Arsha Nagrani, G\\\"ul Varol, Max Bain","submitted_at":"2022-05-17T17:26:23Z","abstract_excerpt":"Our goal in this paper is the adaptation of image-text models for long video retrieval. Recent works have demonstrated state-of-the-art performance in video retrieval by adopting CLIP, effectively hitchhiking on the image-text representation for video tasks. However, there has been limited success in learning temporal aggregation that outperform mean-pooling the image-level representations extracted per frame by CLIP. We find that the simple yet effective baseline of weighted-mean of frame embeddings via query-scoring is a significant improvement above all prior temporal modelling attempts and"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2205.08508","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2205.08508/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2205.08508","created_at":"2026-07-05T04:24:03.800359+00:00"},{"alias_kind":"arxiv_version","alias_value":"2205.08508v1","created_at":"2026-07-05T04:24:03.800359+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2205.08508","created_at":"2026-07-05T04:24:03.800359+00:00"},{"alias_kind":"pith_short_12","alias_value":"IDKVDK63DOKR","created_at":"2026-07-05T04:24:03.800359+00:00"},{"alias_kind":"pith_short_16","alias_value":"IDKVDK63DOKR37PA","created_at":"2026-07-05T04:24:03.800359+00:00"},{"alias_kind":"pith_short_8","alias_value":"IDKVDK63","created_at":"2026-07-05T04:24:03.800359+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21358","citing_title":"LEViL: Label-Efficient Video Learning via Zero-Shot Distillation over VLM-Generated Pseudo-Label Spaces","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2401.07669","citing_title":"SRL-CLIP: Efficient CLIP Video Adaptation via Structured Semantic Role Labels","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2406.05615","citing_title":"Video-Language Understanding: A Survey from Model Architecture, Model Training, and Data Perspectives","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00826","citing_title":"Understanding the Performance Plateau in Text-to-Video Retrieval: A Comprehensive Empirical and Linguistic Analysis","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16487","citing_title":"Geometry-Aware CLIP Retrieval via Local Cross-Modal Alignment and Steering","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IDKVDK63DOKR37PADZNMOLEQ2L","json":"https://pith.science/pith/IDKVDK63DOKR37PADZNMOLEQ2L.json","graph_json":"https://pith.science/api/pith-number/IDKVDK63DOKR37PADZNMOLEQ2L/graph.json","events_json":"https://pith.science/api/pith-number/IDKVDK63DOKR37PADZNMOLEQ2L/events.json","paper":"https://pith.science/paper/IDKVDK63"},"agent_actions":{"view_html":"https://pith.science/pith/IDKVDK63DOKR37PADZNMOLEQ2L","download_json":"https://pith.science/pith/IDKVDK63DOKR37PADZNMOLEQ2L.json","view_paper":"https://pith.science/paper/IDKVDK63","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2205.08508&json=true","fetch_graph":"https://pith.science/api/pith-number/IDKVDK63DOKR37PADZNMOLEQ2L/graph.json","fetch_events":"https://pith.science/api/pith-number/IDKVDK63DOKR37PADZNMOLEQ2L/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IDKVDK63DOKR37PADZNMOLEQ2L/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IDKVDK63DOKR37PADZNMOLEQ2L/action/storage_attestation","attest_author":"https://pith.science/pith/IDKVDK63DOKR37PADZNMOLEQ2L/action/author_attestation","sign_citation":"https://pith.science/pith/IDKVDK63DOKR37PADZNMOLEQ2L/action/citation_signature","submit_replication":"https://pith.science/pith/IDKVDK63DOKR37PADZNMOLEQ2L/action/replication_record"}},"created_at":"2026-07-05T04:24:03.800359+00:00","updated_at":"2026-07-05T04:24:03.800359+00:00"}