← back to paper
arxiv: 2510.18034 · 2 revisions
Can VLMs Unlock Semantic Anomaly Detection? A Framework for Structured Reasoning