TY  - RPRT
TI  - DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models
AU  - Yangfu Li
AU  - Hongjian Zhan
AU  - Jiawei Chen
AU  - Yuning Gong
AU  - Qi Liu
AU  - Yue Lu
PY  - 2026
UR  - https://arxiv.org/abs/2603.03857
ID  - 2603.03857
ER  -