CMDS-AD: Cross-Modal Dual-Stream Decoupling for Few-Shot Anomaly Detection

2026-06-18

AI Summary

This work addresses the challenges of scarce training data, difficult RGB–3D cross-modal alignment, and confusion caused by high-frequency defects in few-shot anomaly detection. To this end, the authors propose a dual-stream decoupling framework that leverages a LoRA-finetuned diffusion model to generate diverse RGB samples. A pretrained diffusion model acts as a nonlinear low-pass filter to extract low-frequency normal structures, forming an auxiliary estimation stream that collaborates with the main stream—preserving genuine mixed-frequency information—to enable precise separation of subtle defects. The method further introduces coordinate-aware hierarchical feature mapping and a multiplicative anomaly scoring mechanism, significantly enhancing cross-modal alignment and noise robustness. Under a 1-shot setting, it achieves state-of-the-art performance, improving I-AUROC by 5.7% and 7.7% and AUPRO by 2.0% and 5.6% on MVTec 3D-AD and EyeCandies, respectively.

Abstract

Few-shot anomaly detection remains challenging due to limited training data. Multi-modal anomaly detection (MAD) offers a viable solution, leveraging 3D geometric cues to enrich 2D RGB representations and compensate for this scarcity. However, existing MAD methods apply spatially uniform feature processing, conflating stable macroscopic structures with high-frequency localized defect signals, exacerbating cross-modal misalignment and inflating false-positive rates. To overcome this, we present CMDS-AD, a Cross-Modal Dual-Stream Anomaly Detection framework. A LoRA-guided diffusion model generates diverse RGB samples to mitigate extreme data scarcity. For 3D normal augmentation, we employ a pre-trained diffusion model as a normal estimator. Crucially, this estimator inherently acts as a non-linear low-pass filter, directly extracting low-frequency normal representations from RGB inputs. This establishes an auxiliary estimated stream of purely low-frequency information, anchoring robust structural templates and assisting the uncompressed real stream, containing coupled high- and low-frequency components, to precisely isolate micro-defects. A Coordinate-Aware Hierarchical Feature Mapper adaptively aligns cross-modal semantics, while a multiplicative scoring mechanism filters modality-specific noise. Under the extreme 1-shot setting, CMDS-AD achieves absolute performance gains of 5.7% (I-AUROC) and 2.0% (AUPRO) on MVTec 3D-AD, alongside 7.7% and 5.6% improvements on EyeCandies, establishing a new state-of-the-art.

Problem

• Few-Shot Anomaly Detection
• Multi-Modal Anomaly Detection
• Cross-Modal Misalignment
• Data Scarcity
• False-Positive Rate

Innovation

• Cross-Modal Dual-Stream
• Few-Shot Anomaly Detection
• Diffusion Model
• Low-Pass Filtering
• Feature Decoupling