EMIT: Enhancing MLLMs for Industrial Anomaly Detection via Difficulty-Aware GRPO

TL;DR

This paper introduces EMIT, a framework that improves multimodal large language models for industrial anomaly detection by using difficulty-aware training strategies and domain-specific data augmentation.

Contribution

EMIT is the first to incorporate difficulty-aware group relative policy optimization for enhancing MLLMs in industrial anomaly detection tasks.

Findings

Achieves 7.77% average performance improvement on MMAD benchmark.

Effectively utilizes GPT-generated descriptions for missing defective images.

Enhances few-shot anomaly detection with soft prompts and contrastive embeddings.

Abstract

Industrial anomaly detection (IAD) plays a crucial role in maintaining the safety and reliability of manufacturing systems. While multimodal large language models (MLLMs) show strong vision-language reasoning abilities, their effectiveness in IAD remains limited without domain-specific adaptation. In this work, we propose EMIT, a unified framework that enhances MLLMs for IAD via difficulty-aware group relative policy optimization (GRPO). EMIT constructs a multi-task IAD dataset and utilizes GPT-generated object text descriptions to compensate for missing defective images. For few-shot anomaly detection, it integrates a soft prompt and heatmap-guided contrastive embeddings derived from patch-level comparisons. To better handle difficult data samples, i.e., cases where the MLLM struggles to generate correct answers, we propose a difficulty-aware GRPO that extends the original GRPO by incorporating a response resampling strategy to ensure the inclusion of correct answers in the sampled responses, as well as an advantage reweighting mechanism to strengthen learning from such difficult data samples. Extensive experiments on the MMAD benchmark demonstrate that EMIT significantly enhances the IAD performance of MLLMs, achieving an average improvement of 7.77\% over the base model (InternVL3-8B) across seven tasks.