The Year in AI Papers
The year/Independent research

Paper 2605.20682

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

Rongbin Tan, Fangfang Lin, Zhenlong Yuan, Min Qiu, Kejin Cui, Mengmeng Wang, Yi Wang, Zijian Song, Zhiyuan Wang, Jiyuan Wang, Yue Wang, Shuhan Song§, et al.

Published
May 2026
Research lab
Independent
Citations
0
GitHub
Not linked

01 In brief

Summary

IndusAgent is a tool-augmented agentic framework for open-vocabulary industrial anomaly detection (IAD).

It addresses limitations of multimodal large language models (MLLMs), such as domain-misaligned reasoning and structural hallucinations, by combining supervised fine-tuning (SFT) with reinforcement learning (RL).

The framework constructs Indus-CoT, a dataset of structured reasoning trajectories with global observations, local patches, and normalcy priors, and uses four tools: dynamic region cropping, normalcy prior retrieval, low-level visual enhancement, and geometric measurement.

An accuracy-gated reward mechanism in GRPO ensures tool use is rewarded only when it improves final diagnostic correctness.

Evaluated on five benchmarks (MVTec-AD, VisA, MPDD, DTD, SDD), IndusAgent achieves state-of-the-art zero-shot performance with an average score of 83.4%, outperforming commercial and open-source baselines, including a 9.3% improvement over the previous best on MVTec.

Ablations confirm the necessity of SFT, RL, and tool augmentation, and the accuracy-gated reward prevents tool abuse while improving anomaly recall.

02 From the paper

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.