Can I read MAMBA4D: Efficient Long-Sequence Point Cloud Video Understanding with Disentangled Spatial-Temporal State Space Models on EtoBox?

MAMBA4D: Efficient Long-Sequence Point Cloud Video Understanding with Disentangled Spatial-Temporal State Space Models by Liu, Jiuming; Han, Jinru; Liu, Lihao; Aviles-Rivero, Angelica I.; Jiang, Chaokang; Liu, Zhe; Wang, Hesheng is a scholarly article available to read on EtoBox.

What is MAMBA4D: Efficient Long-Sequence Point Cloud Video Understanding with Disentangled Spatial-Temporal State Space Models about?

Point cloud videos can faithfully capture real-world spatial geometries and temporal dynamics, which are essential for enabling intelligent agents to understand the dynamically changing world. However, designing an effective 4D backbone remains challenging, mainly due to the irregular and unordered distribution of points and temporal inconsistencies across frames. Also, recent transformer-based 4D backbones commonly suffer from large computational costs due to their quadratic complexity, particularly for long video sequences. To address these challenges, we propose a novel point cloud video understanding backbone purely based on the State Space Models (SSMs). Specifically, we first disentangle space and time in 4D video sequences and then establish the spatio-temporal correlation with our designed Mamba blocks. The Intra-frame Spatial Mamba module is developed to encode locally similar geometric structures within a certain temporal stride. Subsequently, locally correlated tokens are delivered to the Inter-frame Temporal Mamba module, which integrates long-term point features across the entire video with linear complexity. Our proposed Mamba4d achieves competitive performance on the

Author

Liu, Jiuming; Han, Jinru; Liu, Lihao; Aviles-Rivero, Angelica I.; Jiang, Chaokang; Liu, Zhe; Wang, Hesheng

Published

2024

Language

EN