ADAGENT: Anomaly Detection Agent With Multimodal Large Models in Adverse Environments

Multimodal Language Models (MMLMs), such as LLaVA and GPT-4V, have shown zero-shot generalization capabilities for understanding images and text across various domains. However, their effectiveness in open-world visual tasks, particularly anomaly detection under challenging conditions, such as low l...

Ausführliche Beschreibung