🤖 AI Summary
Existing state-space model (SSM)-based methods for 3D human pose estimation rely on manual flattening of 2D pose sequences into unstructured time series, which destroys skeletal topology and conflates spatiotemporal features. To address this, we propose a structure-aware stride-wise SSM architecture: (1) a structure-aware spatiotemporal convolution explicitly models local joint interactions and global pose constraints; (2) a stride-wise multi-scale scanning strategy decouples and jointly captures spatiotemporal dependencies while preserving linear complexity; and (3) a lightweight Mamba backbone enables efficient global modeling. Our approach achieves state-of-the-art performance on benchmarks including Human3.6M, while reducing parameter count by over 40% compared to leading hybrid models. It is the first to unify structural fidelity—preserving anatomical relationships—with computational efficiency in SSM-based 3D pose estimation.
📝 Abstract
Recently, the Mamba architecture based on State Space Models (SSMs) has gained attention in 3D human pose estimation due to its linear complexity and strong global modeling capability. However, existing SSM-based methods typically apply manually designed scan operations to flatten detected 2D pose sequences into purely temporal sequences, either locally or globally. This approach disrupts the inherent spatial structure of human poses and entangles spatial and temporal features, making it difficult to capture complex pose dependencies. To address these limitations, we propose the Skeleton Structure-Aware Stride SSM (SAS-SSM), which first employs a structure-aware spatiotemporal convolution to dynamically capture essential local interactions between joints, and then applies a stride-based scan strategy to construct multi-scale global structural representations. This enables flexible modeling of both local and global pose information while maintaining linear computational complexity. Built upon SAS-SSM, our model SasMamba achieves competitive 3D pose estimation performance with significantly fewer parameters compared to existing hybrid models. The source code is available at https://hucui2022.github.io/sasmamba_proj/.