Passer à la navigation principale Passer à la recherche Passer au contenu principal

Training-Free Uncertainty Guidance for Complex Visual Tasks with MLLMs

  • Technical University of Munich
  • Munich Center for Machine Learning
  • Helmholtz Munich
  • Google Inc.

Résultats de recherche: Le chapitre dans un livre, un rapport, une anthologie ou une collection › Contribution à une conférence › Revue par des pairs

Résumé

Multimodal Large Language Models (MLLMs) often struggle with fine-grained perception, such as identifying small objects in high-resolution images or detecting key moments in long videos. Existing methods typically rely on complex, task-specific fine-tuning, which reduces generalizability and increases system complexity. In this work, we propose an effective, training-free framework that uses an MLLM’s intrinsic uncertainty as proactive guidance. Our core insight is that a model’s uncertainty decreases when provided with relevant visual information. We introduce a unified mechanism that scores candidate visual inputs by response uncertainty, enabling the model to autonomously focus on the most informative data. We apply this simple principle to three challenging visual tasks: Visual Search, Long Video Understanding, and Temporal Grounding, allowing off-the-shelf MLLMs to achieve performance competitive with specialized, fine-tuned systems. Our results demonstrate that leveraging intrinsic uncertainty is a powerful strategy for improving fine-grained multimodal performance. Code is available at https://github.com/ExplainableML/ug-framework.

langue originaleAnglais
titreComputer Vision – ECCV 2026 - 19th European Conference, 2026, Proceedings
rédacteurs en chefPaolo Favaro, Zuzana Kukelova, Atsuto Maki, Anna Rohrbach, Konrad Schindler, Federico Tombari
EditeurSpringer Science and Business Media Deutschland GmbH
Pages606-624
Nombre de pages19
ISBN (imprimé)9783032370198
Les DOIs
étatPublié - 1 janv. 2026
Evénement19th European Conference on Computer Vision, ECCV 2026 - Malmö, Sucde
Durée: 8 sept. 2026 → 12 sept. 2026

Série de publications

NomLecture Notes in Computer Science
Volume17006 LNCS
ISSN (imprimé)0302-9743
ISSN (Electronique)1611-3349

Une conférence

Une conférence19th European Conference on Computer Vision, ECCV 2026
Pays/TerritoireSucde
La villeMalmö
période8/09/26 → 12/09/26

Empreinte digitale

Examiner les sujets de recherche de « Training-Free Uncertainty Guidance for Complex Visual Tasks with MLLMs ». Ensemble, ils forment une empreinte digitale unique.

Contient cette citation