SD-MAD adapts CLIP with shift adapters, an anchor-based loss, and automatic sign-prompt selection to detect and label multiple anomaly types from one or a few medical images.
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
-
SD-MAD: Sign-Driven Few-shot Multi-Anomaly Detection in Medical Images
SD-MAD adapts CLIP with shift adapters, an anchor-based loss, and automatic sign-prompt selection to detect and label multiple anomaly types from one or a few medical images.