ProLearn builds a compact prototype space from paired image-report samples and uses image queries to approximate report semantics, keeping segmentation stable when text is scarce or absent.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Alleviating Textual Reliance in Medical Language-guided Segmentation via Prototype-driven Semantic Approximation
ProLearn builds a compact prototype space from paired image-report samples and uses image queries to approximate report semantics, keeping segmentation stable when text is scarce or absent.