MambaADv2: Evolving Duality-enhanced State Space Model for Unsupervised Anomaly Detection
2026-06-22 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors present MambaADv2, a new method for detecting unusual patterns without needing labeled examples. They build on previous Mamba models to create a system with an encoder and a special decoder that uses Duality-enhanced State Space modules to capture both local details and overall context efficiently. These modules combine different ways of processing data to better reconstruct normal patterns and highlight anomalies. They also introduce a scanning strategy that reduces complexity as the model looks at different feature levels.
anomaly detectionCNNTransformerMamba architecturestate space modelsunsupervised learningfeature pyramidlong-range dependenciesreconstructionrecurrence
Authors
Xiaobin Hu, Haoyang He, Bo Yin, Yu He, Lei Xie, Jiangning Zhang, Yu-Gang Jiang, Shuicheng Yan
Abstract
While recent advancements in anomaly detection have demonstrated the efficacy of CNN- and Transformer-based approaches, these architectures face inherent limitations: CNNs struggle to capture long-range dependencies, whereas Transformers suffer from quadratic computational complexity. Consequently, Mamba-based architectures have attracted considerable attention, as they successfully combine superior long-range dependency modeling with linear computational complexity. By critically rethinking the structural evolution across the Mamba lineage 1-3 series, this paper proposes MambaADv2, a framework tailored for multi-class unsupervised anomaly detection. MambaADv2 comprises a pre-trained encoder and a Mamba-inspired decoder, equipped with Duality-enhanced State Space (DSS) modules across multiple scales. The proposed DSS module effectively models both global dependencies and local representations by integrating parallel-cascaded Hybrid State Space (HSS) blocks and frequency-enhanced convolution operations. The structure of the Hybrid State Space (HSS) block is tailored by following the SSD-based Mamba lineage and incorporating Mamba3-style position-aware state-space modeling, leveraging the dual computational paths of linear recurrence and parallel matrix formulation to model local continuity and global contextual comparison, thereby better serving the core anomaly detection objective of precisely reconstructing normal representations while magnifying anomalous deviations. Additionally, we propose a semantics-adaptive progressive scanning strategy that decays scanning complexity along the feature pyramid.