English(EN) LDDR: Linear-DPP Dynamic-Resolution Frame Sampling for Video MLLMs

LDDR框架优化多模态大语言模型的视频帧采样

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-19 04:00

研究人员推出了一种新颖的框架LDDR，用于从视频中采样帧以优化多模态大语言模型（MLLMs）的输入。LDDR采用查询感知的行列式点过程（DPP）来高效选择帧，比标准DPP方法快3倍。该框架还包含一个组DPP重要性得分，用于动态分配token预算，优先考虑信息量大的帧并缩减不太相关的帧。LDDR在视频基准测试中已显示出显著的收益，在低预算设置下性能提高高达2.5个点，在高预算场景下提高1.6个点，覆盖了各种MLLM家族。 AI

影响通过优化帧选择和token分配，提高了MLLMs在视频理解方面的效率和性能。

排序理由关于一种用于MLLMs视频帧采样新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.CV 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 8 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 关于一种用于MLLMs视频帧采样新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, infra; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Same-day; Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.CV TIER_1 English(EN) · Jingfeng Chen, Jiawen Qian, Wendi Deng, Yinuo Guo, Jiaqi Yu, Sicong Leng, Raghuveer Thirukovalluru, Bhuwan Dhingra · 2026-10-19 04:00

LDDR：面向视频多模态大模型的线性DPP动态分辨率帧采样

arXiv:2605.11477v2 Announce Type: replace Abstract: Multimodal large language models, when used for long video understanding, require selecting an informative frame set from massive and redundant candidates. Limited by the model's visual-token budget and reasoning ability, frame …

本文如何被排名

报道来源 [1]

LDDR：面向视频多模态大模型的线性DPP动态分辨率帧采样

相关实体

相关话题