{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:2TKZVMYPCQ7GATY3WQJKZGNPYH","short_pith_number":"pith:2TKZVMYP","schema_version":"1.0","canonical_sha256":"d4d59ab30f143e604f1bb412ac99afc1d8106461eba19708354923ffd4b87809","source":{"kind":"arxiv","id":"2410.20252","version":1},"attestation_state":"computed","paper":{"title":"Adaptive Video Understanding Agent: Enhancing efficiency with dynamic frame sampling and feedback-driven reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Aram Galstyan, Bhavana Ganesh, Goeric Huybrechts, Sravan Bodapati, Sullam Jeoung","submitted_at":"2024-10-26T19:01:06Z","abstract_excerpt":"Understanding long-form video content presents significant challenges due to its temporal complexity and the substantial computational resources required. In this work, we propose an agent-based approach to enhance both the efficiency and effectiveness of long-form video understanding by utilizing large language models (LLMs) and their tool-harnessing ability. A key aspect of our method is query-adaptive frame sampling, which leverages the reasoning capabilities of LLMs to process only the most relevant frames in real-time, and addresses an important limitation of existing methods which typica"},"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":"2410.20252","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-10-26T19:01:06Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"470209bf40ec04536c7ffe671954c4415c3731bca363b65dc525a6bed5dfbb17","abstract_canon_sha256":"653732a39f5f0eb40de4cf4b67ed9ef846ff352bb277f8867f8c97999672e9ce"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:26:41.471252Z","signature_b64":"dlIte3VgnyxBFqoY0F21qKxisB6qXJ2XNnfSrEZ5JB6nkeEX6MsY8gJGuopm/ek6lE+PYyvwskpvMaFm63S2Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d4d59ab30f143e604f1bb412ac99afc1d8106461eba19708354923ffd4b87809","last_reissued_at":"2026-07-05T09:26:41.470737Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:26:41.470737Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Adaptive Video Understanding Agent: Enhancing efficiency with dynamic frame sampling and feedback-driven reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Aram Galstyan, Bhavana Ganesh, Goeric Huybrechts, Sravan Bodapati, Sullam Jeoung","submitted_at":"2024-10-26T19:01:06Z","abstract_excerpt":"Understanding long-form video content presents significant challenges due to its temporal complexity and the substantial computational resources required. In this work, we propose an agent-based approach to enhance both the efficiency and effectiveness of long-form video understanding by utilizing large language models (LLMs) and their tool-harnessing ability. A key aspect of our method is query-adaptive frame sampling, which leverages the reasoning capabilities of LLMs to process only the most relevant frames in real-time, and addresses an important limitation of existing methods which typica"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.20252","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/2410.20252/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":"2410.20252","created_at":"2026-07-05T09:26:41.470799+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.20252v1","created_at":"2026-07-05T09:26:41.470799+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.20252","created_at":"2026-07-05T09:26:41.470799+00:00"},{"alias_kind":"pith_short_12","alias_value":"2TKZVMYPCQ7G","created_at":"2026-07-05T09:26:41.470799+00:00"},{"alias_kind":"pith_short_16","alias_value":"2TKZVMYPCQ7GATY3","created_at":"2026-07-05T09:26:41.470799+00:00"},{"alias_kind":"pith_short_8","alias_value":"2TKZVMYP","created_at":"2026-07-05T09:26:41.470799+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22804","citing_title":"CoVStream: Edge-Cloud Collaboration for Understanding of Long Video Streams","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07433","citing_title":"Watch, Remember, Reason: Human-View Video Understanding with MLLMs","ref_index":140,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29445","citing_title":"Bridging VideoQA and Video-Guided Agentic Tasks via Generalized Keyframe Extraction","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10762","citing_title":"GridProbe: Posterior-Probing for Adaptive Test-Time Compute in Long-Video VLMs","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2TKZVMYPCQ7GATY3WQJKZGNPYH","json":"https://pith.science/pith/2TKZVMYPCQ7GATY3WQJKZGNPYH.json","graph_json":"https://pith.science/api/pith-number/2TKZVMYPCQ7GATY3WQJKZGNPYH/graph.json","events_json":"https://pith.science/api/pith-number/2TKZVMYPCQ7GATY3WQJKZGNPYH/events.json","paper":"https://pith.science/paper/2TKZVMYP"},"agent_actions":{"view_html":"https://pith.science/pith/2TKZVMYPCQ7GATY3WQJKZGNPYH","download_json":"https://pith.science/pith/2TKZVMYPCQ7GATY3WQJKZGNPYH.json","view_paper":"https://pith.science/paper/2TKZVMYP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.20252&json=true","fetch_graph":"https://pith.science/api/pith-number/2TKZVMYPCQ7GATY3WQJKZGNPYH/graph.json","fetch_events":"https://pith.science/api/pith-number/2TKZVMYPCQ7GATY3WQJKZGNPYH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2TKZVMYPCQ7GATY3WQJKZGNPYH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2TKZVMYPCQ7GATY3WQJKZGNPYH/action/storage_attestation","attest_author":"https://pith.science/pith/2TKZVMYPCQ7GATY3WQJKZGNPYH/action/author_attestation","sign_citation":"https://pith.science/pith/2TKZVMYPCQ7GATY3WQJKZGNPYH/action/citation_signature","submit_replication":"https://pith.science/pith/2TKZVMYPCQ7GATY3WQJKZGNPYH/action/replication_record"}},"created_at":"2026-07-05T09:26:41.470799+00:00","updated_at":"2026-07-05T09:26:41.470799+00:00"}