PointLAMA: Latent Attention meets Mamba for Efficient Point Cloud Pretraining

2025/07/23 by Xuanyu Lin, Lin, Xuanyu, Zeng, Xiaona +4 · 1 citation
Computer Science · Earth and Planetary Sciences · Engineering · #3D Shape Modeling and Analysis #3D Surveying and Cultural Heritage #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Image Processing and 3D Reconstruction

paper · pdf · doi:10.48550/arxiv.2507.17296

openalex publication_date 2025/07/23 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Mamba has recently gained widespread attention as a backbone model for point cloud modeling, leveraging a state-space architecture that enables efficient global sequence modeling with linear complexity. However, its lack of local inductive bias limits its capacity to capture fine-grained geometric structures in 3D data. To address this limitation, we propose PointLAMA, a point cloud pretraining framework that combines task-aware point cloud serialization, a hybrid encoder with integrated Latent Attention and Mamba blocks, and a conditional diffusion mechanism built upon the Mamba backbone. Specifically, the task-aware point cloud serialization employs Hilbert/Trans-Hilbert space-filling curves and axis-wise sorting to structurally align point tokens for classification and segmentation tasks, respectively. Our lightweight Latent Attention block features a Point-wise Multi-head Latent Attention (PMLA) module, which is specifically designed to align with the Mamba architecture by leveraging the shared latent space characteristics of PMLA and Mamba. This enables enhanced local context modeling while preserving overall efficiency. To further enhance representation learning, we incorporate a conditional diffusion mechanism during pretraining, which denoises perturbed feature sequences without relying on explicit point-wise reconstruction. Experimental results demonstrate that PointLAMA achieves competitive performance on multiple benchmark datasets with minimal parameter count and FLOPs, validating its effectiveness for efficient point cloud pretraining.

Citations

• Mamba: Linear-Time Sequence Modeling with Selective State Spaces
• Point Cloud Pre-training with Diffusion Models
• PointGPT: Auto-regressively Generative Pre-training from Point Clouds
• Instance-aware Dynamic Prompt Tuning for Pre-trained Point Cloud Models
• Joint-MAE: 2D-3D Joint Masked Autoencoders for 3D Point Cloud Pre-training
• Contrast with Reconstruct: Contrastive 3D Representation Learning Guided by Generative Pretraining
• Autoencoders as Cross-Modal Teachers: Can Pretrained 2D Image Transformers Help 3D Representation Learning?
• How to Train Your HiPPO: State Space Models with Generalized Orthogonal Basis Projections
• Diagonal State Spaces are as Effective as Structured State Spaces
• Masked Discrimination for Self-Supervised Learning on Point Clouds
• Point-BERT: Pre-training 3D Point Cloud Transformers with Masked Point Modeling
• Efficiently Modeling Long Sequences with Structured State Spaces
• 3D Shape Generation and Completion through Point-Voxel Diffusion
• Perceiver: General Perception with Iterative Attention
• Transformer in Transformer
• PCT: Point cloud transformer
• Self-supervised Learning of Point Clouds via Orientation Estimation
• Denoising Diffusion Probabilistic Models
• Revisiting Point Cloud Classification: A New Benchmark Dataset and\n Classification Model on Real-World Data
• Dynamic Graph CNN for Learning on Point Clouds
• Dynamic Graph CNN for Learning on Point Clouds
• PointCNN: Convolution On X-Transformed Points
• Attention Is All You Need
• PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space
• A Point Set Generation Network for 3D Object Reconstruction from a Single Image
• PointNet: Deep Learning on Point Sets for 3D Classification and\n Segmentation
• A scalable active framework for region annotation in 3D shape collections
• Deep Unsupervised Learning using Nonequilibrium Thermodynamics
• The Earth Mover's Distance as a Metric for Image Retrieval

Cited by

• Dual-Branch Center-Surrounding Contrast: Rethinking Contrastive Learning for 3D Point Clouds

Related

• Point Cloud Mamba: Point Cloud Learning via State Space Model
• PMA: Towards Parameter-Efficient Point Cloud Understanding via Point Mamba Adapter
• StruMamba3D: Exploring Structural Mamba for Self-supervised Point Cloud Representation Learning
• Serialized Point Mamba: A Serialized Point Cloud Mamba Segmentation Model
• HydraMamba: Multi-Head State Space Model for Global Point Cloud Learning