{"paper":{"title":"SRL-CLIP: Efficient CLIP Video Adaptation via Structured Semantic Role Labels","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"Structured semantic role label captions let CLIP adapt to video tasks with only 23k pairs.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Darshan Singh, Makarand Tapaswi, Zeeshan Khan","submitted_at":"2024-01-15T13:27:34Z","abstract_excerpt":"Adapting CLIP for videos has gained popularity due to its semantic and rich representation. While CLIP is a good starting point, it typically undergoes post-pretraining (contrastive finetuning) on large video narration or caption datasets (e.g. HowTo100M, WebVid2.5M). However, such narrations or captions often lack comprehensive information needed to represent a video holistically. As the learning signal from text is sparse, the visual learning is inefficient and adaptation requires millions of samples to post-pretrain. In this work, we ask: is it possible to efficiently adapt CLIP for general"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"simple contrastive finetuning on a mere 23k video-caption pairs is adequate to learn powerful, transferable representations applicable across a diverse range of video understanding tasks that require varying levels of perceptual granularity","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that rule-based captions generated from SRL annotations supply a sufficiently rich and holistic learning signal compared with the sparse narrations found in large-scale video datasets.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"SRL-CLIP uses rule-based captions derived from semantic role labels to adapt CLIP via contrastive fine-tuning on 23k pairs, matching or exceeding larger models trained on far more data across video tasks.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Structured semantic role label captions let CLIP adapt to video tasks with only 23k pairs.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"ba2ebc07fb3a874cbe15b8e8c4fd5e320f52630e60c333fc2363c5cc1484a372"},"source":{"id":"2401.07669","kind":"arxiv","version":3},"verdict":{"id":"0874eff8-3e60-4cab-a376-c3686baea2de","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-24T04:44:46.609542Z","strongest_claim":"simple contrastive finetuning on a mere 23k video-caption pairs is adequate to learn powerful, transferable representations applicable across a diverse range of video understanding tasks that require varying levels of perceptual granularity","one_line_summary":"SRL-CLIP uses rule-based captions derived from semantic role labels to adapt CLIP via contrastive fine-tuning on 23k pairs, matching or exceeding larger models trained on far more data across video tasks.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that rule-based captions generated from SRL annotations supply a sufficiently rich and holistic learning signal compared with the sparse narrations found in large-scale video datasets.","pith_extraction_headline":"Structured semantic role label captions let CLIP adapt to video tasks with only 23k pairs."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2401.07669/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"}