AI News HubLIVE
原文

Not All Modalities Are Equal: Instruction-Aware Gating for Multimodal Videos

Pre-trained video LLMs struggle with auxiliary streams like audio or depth maps due to modality interference. UniMVU uses instruction-aware dynamic gating at two levels (inner-modality and modality-level) to adaptively balance importance, achieving gains up to 13.5 CIDEr across six benchmarks and aligning with human-interpretable relevance.

Article intelligence

EngineersAdvanced

Key points

  • UniMVU introduces instruction-aware gating with inner-modality gates (emphasizing salient regions) and modality-level gates (re-weighting streams), conditioned on text instructions.
  • The framework combines cross-modal self-attention with instruction-driven gating modules and a fast-to-slow fusion for time-aligned streams to reduce redundancy.
  • Consistent improvements over static-fusion baselines on six benchmarks (AVQA, AVSD, Music-AVQA, ScanQA, SQA3D, MVBench), with up to 13.5 CIDEr gain.

Why it matters

This matters because uniMVU introduces instruction-aware gating with inner-modality gates (emphasizing salient regions) and modality-level gates (re-weighting streams), conditioned on text instructions.

Technical impact

May affect model selection, inference cost, product capability, and evaluation benchmarks.

[2605.26232] Not All Modalities Are Equal: Instruction-Aware Gating for Multimodal Videos

[Submitted on 25 May 2026]

Title:Not All Modalities Are Equal: Instruction-Aware Gating for Multimodal Videos

View a PDF of the paper titled Not All Modalities Are Equal: Instruction-Aware Gating for Multimodal Videos, by Bonan Ding and 6 other authors

View PDF HTML (experimental)

Abstract:Pre-trained video large language models excel at visual reasoning. However, they struggle when videos arrive with auxiliary streams, such as audio, depth map, or dense temporal evidence. In such a scenario, uniform fusion induces modality interference, allowing irrelevant channels to distract the model. To address this issue, we present a unified multimodal video understanding framework, named UniMVU, that performs instruction-aware fusion across video, audio, depth map, or any other modality inputs via two levels of dynamic gating: inner-modality gates emphasize salient regions within each modality, whereas modality-level gates re-weight whole streams; both are conditioned on the text instruction to adaptively balance modality importance. Our UniMVU combines cross-modal self-attention with instruction-driven inner-modality gating module and a modality-level gating module with control token; for time-aligned streams we further adopt a fast-to-slow fusion scheme that reduces redundancy. Across six benchmarks (AVQA, AVSD, Music-AVQA, ScanQA, SQA3D and MVBench), our UniMVU achieves consistent gains over static-fusion baselines achieving gains as high as 13.5 in terms of CIDEr metric. Further, our analysis shows that the gating mechanism aligns with the human-interpretable modality relevance, and ablations show the contributions of inner-modality and modality-level gating. Our UniMVU provides a simple, unified recipe for instruction-aware multimodal video understanding that scales to diverse modalities without hand-crafted fusion rules.

Comments: 19 pages, 8 figures, 7 tables, preprint

Subjects:

Computer Vision and Pattern Recognition (cs.CV)

Cite as: arXiv:2605.26232 [cs.CV]

(or arXiv:2605.26232v1 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2605.26232

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Bonan Ding [view email] [v1] Mon, 25 May 2026 18:02:20 UTC (11,190 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Not All Modalities Are Equal: Instruction-Aware Gating for Multimodal Videos, by Bonan Ding and 6 other authors

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.CV

new | recent | 2026-05

Change to browse by:

cs

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)