Foveated Reasoning: Stateful, Action-based Visual Focusing for Vision-Language Models

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Min, Juhong, Valkov, Lazar, Petsiuk, Vitali, Souri, Hossein, Mohan, Deen Dayal
Format:	Preprint
Published:	2026
Subjects:	Computer Vision and Pattern Recognition
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Internet

https://arxiv.org/abs/2604.21079