MAU-GPT: Enhancing Multi-type Industrial Anomaly Understanding via Anomaly-aware and Generalist Experts Adaptation

TL;DR

MAU-GPT is a multimodal large model designed for industrial anomaly understanding, leveraging a new dataset and a novel adaptation mechanism to improve generalization and reasoning across diverse defect types.

Contribution

The paper introduces MAU-GPT with a novel AMoE-LoRA mechanism and a comprehensive industrial anomaly dataset, advancing multi-type anomaly detection and reasoning capabilities.

Findings

MAU-GPT outperforms previous methods across multiple industrial domains.

The AMoE-LoRA mechanism enhances anomaly understanding and reasoning.

The MAU-Set dataset enables comprehensive evaluation of anomaly detection models.

Abstract

As industrial manufacturing scales, automating fine-grained product image analysis has become critical for quality control. However, existing approaches are hindered by limited dataset coverage and poor model generalization across diverse and complex anomaly patterns. To address these challenges, we introduce MAU-Set, a comprehensive dataset for Multi-type industrial Anomaly Understanding. It spans multiple industrial domains and features a hierarchical task structure, ranging from binary classification to complex reasoning. Alongside this dataset, we establish a rigorous evaluation protocol to facilitate fair and comprehensive model assessment. Building upon this foundation, we further present MAU-GPT, a domain-adapted multimodal large model specifically designed for industrial anomaly understanding. It incorporates a novel AMoE-LoRA mechanism that unifies anomaly-aware and generalist experts adaptation, enhancing both understanding and reasoning across diverse defect classes. Extensive experiments show that MAU-GPT consistently outperforms prior state-of-the-art methods across all domains, demonstrating strong potential for scalable and automated industrial inspection.