A dual-side evidence-injection method using ROI-guided modulation and semantic token mapping improves medical MLLM close-ended accuracy by up to 6% and cuts open-ended hallucinations by 35% across 5 datasets.
In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Synergistic Perception-Reasoning Governance: Grounding Medical MLLMs with Verifiable Anatomical Evidence
A dual-side evidence-injection method using ROI-guided modulation and semantic token mapping improves medical MLLM close-ended accuracy by up to 6% and cuts open-ended hallucinations by 35% across 5 datasets.