IndusAgent: Reinforcing Open-Vocabulary Industrial Anomaly Detection with Agentic Tools

2026-05-19

AI Summary

This work addresses the performance limitations of multimodal large language models in open-vocabulary industrial anomaly detection, which stem from domain mismatch and structural hallucination. To overcome these challenges, we propose the first tool-augmented agent framework tailored for this task. The framework dynamically invokes external tools—such as region cropping, high-frequency feature enhancement, and prior retrieval—to actively resolve visual ambiguities and detect subtle anomalies. A gated reinforcement learning objective jointly optimizes classification, localization, anomaly-type reasoning, and tool usage efficiency. To facilitate learning of industrial inspection strategies, we introduce Indus-CoT, a structured chain-of-thought dataset. Our method achieves zero-shot state-of-the-art performance across five benchmarks, including MVTec-AD and VisA, significantly enhancing model robustness and generalization.

Abstract

Multimodal large language models (MLLMs) have shown remarkable capability in bridging visual perception and textual reasoning, enabling zero-shot understanding across diverse industrial scenarios. However, their performance in open-vocabulary industrial anomaly detection (IAD) is often limited by domain-misaligned reasoning and hallucinated structural inferences. To address these challenges, we propose \textbf{IndusAgent}, a tool-augmented agentic framework for open-vocabulary IAD. Specifically, we first construct \textbf{Indus-CoT}, a structured dataset that integrates global visual observations, high-resolution local patches, and expert normalcy priors, providing supervision for fine-tuning the model on rigorous industrial inspection trajectories. Building on this, IndusAgent dynamically orchestrates a set of external tools, including dynamic region cropping, high-frequency feature enhancement, and prior retrieval, thus enabling the agent to actively resolve visual ambiguities and disentangle subtle anomalies. Furthermore, we introduce a gated reinforcement learning objective that jointly optimizes anomaly classification, localization accuracy, anomaly type reasoning, and efficient tool usage, ensuring that tool invocation occurs only when beneficial. Extensive evaluations on five industrial anomaly benchmarks, including MVTec-AD, VisA, MPDD, DTD, and SDD, demonstrate that IndusAgent achieves state-of-the-art zero-shot performance among all existing methods, validating our robustness and generalization capacity.

Problem

• open-vocabulary industrial anomaly detection
• domain-misaligned reasoning
• hallucinated structural inferences
• multimodal large language models
• industrial anomaly detection

Innovation

• tool-augmented agent
• open-vocabulary anomaly detection
• multimodal reasoning
• gated reinforcement learning
• industrial inspection