DDL enables test-time refinement of frozen LVLMs for abnormality grounding in rare diseases, improving mAP@75 by up to 105% over baselines on a 281-pathology dataset.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Dynamic Decision Learning: Test-Time Evolution for Abnormality Grounding in Rare Diseases
DDL enables test-time refinement of frozen LVLMs for abnormality grounding in rare diseases, improving mAP@75 by up to 105% over baselines on a 281-pathology dataset.