Search papers, labs, and topics across Lattice.
This paper introduces SAGE, a Vision-Language Model (VLM) framework designed for industrial anomaly detection, addressing the limitations of existing VLMs in this domain. SAGE incorporates Self-Guided Fact Enhancement (SFE) to inject domain-specific knowledge and Entropy-aware Direct Preference Optimization (E-DPO) to align model outputs with expert preferences. Experiments on industrial anomaly datasets demonstrate SAGE's superior performance in zero-shot and one-shot settings, validated using a newly proposed Multiscale Logical Evaluation (MLE) framework.
VLMs can now excel at industrial anomaly detection by injecting domain-specific facts and aligning with expert preferences, achieving state-of-the-art zero-shot and one-shot performance.
While Vision-Language Models (VLMs) have shown promising progress in general multimodal tasks, they often struggle with industrial anomaly detection and reasoning, particularly in delivering interpretable explanations and generalizing to unseen categories. This limitation stems from the inherently domain-specific nature of anomaly detection, which hinders the applicability of existing VLMs in industrial scenarios that require precise, structured, and context-aware analysis. To address these challenges, we propose SAGE, a VLM-based framework that enhances anomaly reasoning through Self-Guided Fact Enhancement (SFE) and Entropy-aware Direct Preference Optimization (E-DPO). SFE integrates domain-specific knowledge into visual reasoning via fact extraction and fusion, while E-DPO aligns model outputs with expert preferences using entropy-aware optimization. Additionally, we introduce AD-PL, a preference-optimized dataset tailored for industrial anomaly reasoning, consisting of 28,415 question-answering instances with expert-ranked responses. To evaluate anomaly reasoning models, we develop Multiscale Logical Evaluation (MLE), a quantitative framework analyzing model logic and consistency. SAGE demonstrates superior performance on industrial anomaly datasets under zero-shot and one-shot settings. The code, model, and dataset are available at https://github.com/amoreZgx1n/SAGE.
Tsinghua AI