DM3D: Deformable Mamba via Offset-Guided Gaussian Sequencing for Point Cloud Understanding
By: Bin Liu, Chunyang Wang, Xuelian Liu
Potential Business Impact:
Helps computers understand 3D shapes better.
State Space Models (SSMs) demonstrate significant potential for long-sequence modeling, but their reliance on input order conflicts with the irregular nature of point clouds. Existing approaches often rely on predefined serialization strategies, which cannot adjust based on diverse geometric structures. To overcome this limitation, we propose \textbf{DM3D}, a deformable Mamba architecture for point cloud understanding. Specifically, DM3D introduces an offset-guided Gaussian sequencing mechanism that unifies local resampling and global reordering within a deformable scan. The Gaussian-based KNN Resampling (GKR) enhances structural awareness by adaptively reorganizing neighboring points, while the Gaussian-based Differentiable Reordering (GDR) enables end-to-end optimization of serialization order. Furthermore, a Tri-Path Frequency Fusion module enhances feature complementarity and reduces aliasing. Together, these components enable structure-adaptive serialization of point clouds. Extensive experiments on benchmark datasets show that DM3D achieves state-of-the-art performance in classification, few-shot learning, and part segmentation, demonstrating that adaptive serialization effectively unlocks the potential of SSMs for point cloud understanding.
Similar Papers
DAPointMamba: Domain Adaptive Point Mamba for Point Cloud Completion
CV and Pattern Recognition
Fixes 3D shape gaps in computer vision.
DM-SegNet: Dual-Mamba Architecture for 3D Medical Image Segmentation with Global Context Modeling
Image and Video Processing
Improves medical scans for better diagnoses.
UniMamba: Unified Spatial-Channel Representation Learning with Group-Efficient Mamba for LiDAR-based 3D Object Detection
CV and Pattern Recognition
Helps self-driving cars see in 3D better.