The Segment Anything Model (SAM) is a foundation model for prompt-driven image segmentation tasks. Manual prompting increases the burden, leading to subpar performance in medical imaging. UR-SAM proposes uncertainty rectification to improve reliability without manual prompting. Experiments show up to 13.8% improvement in segmentation performance.
toiselle kielelle
lähdeaineistosta
arxiv.org
Tärkeimmät oivallukset
by Yichi Zhang,... klo arxiv.org 03-19-2024
https://arxiv.org/pdf/2311.10529.pdfSyvällisempiä Kysymyksiä