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

• URL: http://arxiv.org/abs/2405.14338v2
• Date: Sun, 24 Nov 2024 04:26:03 GMT
• Title: MAMBA4D: Efficient Long-Sequence Point Cloud Video Understanding with Disentangled Spatial-Temporal State Space Models
• Authors: Jiuming Liu, Jinru Han, Lihao Liu, Angelica I. Aviles-Rivero, Chaokang Jiang, Zhe Liu, Hesheng Wang,
• Abstract summary: We propose a novel point cloud understanding backbone based on the State Space Models (SSMs) Specifically, we first disentangle space and time in 4D video sequences and then establish the spatial-temporal correlation with our designed Mamba blocks. Our method has a significant efficiency improvement with 87.5% GPU memory reduction and 5.36 times speed-up.
• Score: 14.024240637175216
• License:
• Abstract: 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 MSR-Action3D action recognition (+10.4% accuracy), HOI4D action segmentation (+0.7 F1 Score), and Synthia4D semantic segmentation (+0.19 mIoU) datasets. Especially, for long video sequences, our method has a significant efficiency improvement with 87.5% GPU memory reduction and 5.36 times speed-up.

Related papers

• Seeing World Dynamics in a Nutshell [132.79736435144403]
  NutWorld is a framework that transforms monocular videos into dynamic 3D representations in a single forward pass. We demonstrate that NutWorld achieves high-fidelity video reconstruction quality while enabling downstream applications in real-time.
  arXiv  Detail & Related papers  (2025-02-05T18:59:52Z)
• 4D Gaussian Splatting: Modeling Dynamic Scenes with Native 4D Primitives [116.2042238179433]
  In this paper, we frame dynamic scenes as unconstrained 4D volume learning problems. We represent a target dynamic scene using a collection of 4D Gaussian primitives with explicit geometry and appearance features. This approach can capture relevant information in space and time by fitting the underlying photorealistic-temporal volume. Notably, our 4DGS model is the first solution that supports real-time rendering of high-resolution, novel views for complex dynamic scenes.
  arXiv  Detail & Related papers  (2024-12-30T05:30:26Z)
• STNMamba: Mamba-based Spatial-Temporal Normality Learning for Video Anomaly Detection [48.997518615379995]
  Video anomaly detection (VAD) has been extensively researched due to its potential for intelligent video systems. Most existing methods based on CNNs and transformers still suffer from substantial computational burdens. We propose a lightweight and effective Mamba-based network named STNMamba to enhance the learning of spatial-temporal normality.
  arXiv  Detail & Related papers  (2024-12-28T08:49:23Z)
• When Spatial meets Temporal in Action Recognition [34.53091498930863]
  We introduce the Temporal Integration and Motion Enhancement (TIME) layer, a novel preprocessing technique designed to incorporate temporal information. The TIME layer generates new video frames by rearranging the original sequence, preserving temporal order while embedding $N2$ temporally evolving frames into a single spatial grid. Our experiments show that the TIME layer enhances recognition accuracy, offering valuable insights for video processing tasks.
  arXiv  Detail & Related papers  (2024-11-22T16:39:45Z)
• Flow4D: Leveraging 4D Voxel Network for LiDAR Scene Flow Estimation [20.904903264632733]
  Flow4D temporally fuses multiple point clouds after the 3D intra-voxel feature encoder. Spatio-Temporal De Blockcomposition (STDB) combines 3D and 1D convolutions instead of using heavy 4D convolutions. Flow4D achieves a 45.9% higher performance compared to the state-of-the-art while running in real-time.
  arXiv  Detail & Related papers  (2024-07-10T18:55:43Z)
• NeRFPlayer: A Streamable Dynamic Scene Representation with Decomposed Neural Radiance Fields [99.57774680640581]
  We present an efficient framework capable of fast reconstruction, compact modeling, and streamable rendering. We propose to decompose the 4D space according to temporal characteristics. Points in the 4D space are associated with probabilities belonging to three categories: static, deforming, and new areas.
  arXiv  Detail & Related papers  (2022-10-28T07:11:05Z)
• Temporally Consistent Transformers for Video Generation [80.45230642225913]
  To generate accurate videos, algorithms have to understand the spatial and temporal dependencies in the world. No established benchmarks on complex data exist for rigorously evaluating video generation with long temporal dependencies. We introduce the Temporally Consistent Transformer (TECO), a generative model that substantially improves long-term consistency while also reducing sampling time.
  arXiv  Detail & Related papers  (2022-10-05T17:15:10Z)
• Learning Dynamic View Synthesis With Few RGBD Cameras [60.36357774688289]
  We propose to utilize RGBD cameras to synthesize free-viewpoint videos of dynamic indoor scenes. We generate point clouds from RGBD frames and then render them into free-viewpoint videos via a neural feature. We introduce a simple Regional Depth-Inpainting module that adaptively inpaints missing depth values to render complete novel views.
  arXiv  Detail & Related papers  (2022-04-22T03:17:35Z)
• STH: Spatio-Temporal Hybrid Convolution for Efficient Action Recognition [39.58542259261567]
  We present a novel S-Temporal Hybrid Network (STH) which simultaneously encodes spatial and temporal video information with a small parameter. Such a design enables efficient-temporal modeling and maintains a small model scale. STH enjoys performance superiority over 3D CNNs while maintaining an even smaller parameter cost than 2D CNNs.
  arXiv  Detail & Related papers  (2020-03-18T04:46:30Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.