Jul 2026
Global Logic and Local Search: Dual-Stream Multimodal In-Context Learning for Verifiable Industrial Anomaly Detection
This work presents GLLS, a training-free framework for reference-guided multimodal in-context verification that combines a Global&Logic Stream, where SAM 3 extracts partially checkable visual facts, with a Fine-Grained&Actions Stream, where MCTS selects local evidence crops under a fixed budget.
R. Deng, Yundi Hu, Yi-Ming Zhong et al.
· arXiv.org · 0 citations