Pith. sign in

REVIEW 4 cited by

Threading Keyframe with Narratives: MLLMs as Strong Long Video Comprehenders

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2505.24158 v1 pith:K4JYYSFU submitted 2025-05-30 cs.CV

Threading Keyframe with Narratives: MLLMs as Strong Long Video Comprehenders

classification cs.CV
keywords mllmsnar-kfcnarrativesvideokeyframelongcomputationalefficient
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Employing Multimodal Large Language Models (MLLMs) for long video understanding remains a challenging problem due to the dilemma between the substantial number of video frames (i.e., visual tokens) versus the limited context length of language models. Traditional uniform sampling often leads to selection of irrelevant content, while post-training MLLMs on thousands of frames imposes a substantial computational burden. In this paper, we propose threading keyframes with narratives (Nar-KFC), a plug-and-play module to facilitate effective and efficient long video perception. Nar-KFC generally involves two collaborative steps. First, we formulate the keyframe selection process as an integer quadratic programming problem, jointly optimizing query-relevance and frame-diversity. To avoid its computational complexity, a customized greedy search strategy is designed as an efficient alternative. Second, to mitigate the temporal discontinuity caused by sparse keyframe sampling, we further introduce interleaved textual narratives generated from non-keyframes using off-the-shelf captioners. These narratives are inserted between keyframes based on their true temporal order, forming a coherent and compact representation. Nar-KFC thus serves as a temporal- and content-aware compression strategy that complements visual and textual modalities. Experimental results on multiple long-video benchmarks demonstrate that Nar-KFC significantly improves the performance of popular MLLMs. Code will be made publicly available.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. CREST: Curvature-Regulated Event-Centric Sampling for Efficient Long-Video Understanding

    cs.CV 2026-05 unverdicted novelty 7.0

    CATS uses temporal curvature of query-frame relevance to select informative frames, achieving 93-95% of heavy multi-stage accuracy at 3-4% of the preprocessing cost on long-video benchmarks.

  2. CREST: Curvature-Regulated Event-Centric Sampling for Efficient Long-Video Understanding

    cs.CV 2026-05 conditional novelty 6.0

    CREST selects video frames using curvature-adaptive non-maximum suppression on CLIP relevance scores, beating AKS by ~0.5% on two long-video QA benchmarks while using a fraction of MIRA's preprocessing cost.

  3. Training-free Uncertainty Guidance for Complex Visual Tasks with MLLMs

    cs.CV 2025-10 conditional novelty 6.0

    Selecting the visual input that minimizes an MLLM's output entropy (or maximizes its yes/no confidence) improves fine-grained visual search, long-video QA, and temporal grounding without any training.

  4. CREST: Curvature-Regulated Event-Centric Sampling for Efficient Long-Video Understanding

    cs.CV 2026-05 unverdicted novelty 4.0

    CREST uses local curvature of query-frame relevance over time to select informative frames, outperforming a lightweight baseline and approaching a costly pipeline at far lower preprocessing cost on long-video benchmarks.