PointLAM: Local Attentive Mamba for Efficient Point-based 3D Object Detection
作者: Xuanming Shang, Weijia Zhang, Chao Ma
分类: cs.CV
发布日期: 2026-09-18
备注: Accepted to ECCV 2026
🔗 代码/项目: PROJECT_PAGE
💡 一句话要点
提出PointLAM以解决点云3D目标检测的效率与精度问题
🎯 匹配领域: 支柱二:RL算法与架构 (RL & Architecture)
关键词: 3D目标检测 点云处理 深度学习 自动驾驶 高效算法
📋 核心要点
- 现有的点云3D目标检测方法在效率与几何精度之间存在权衡,导致计算瓶颈。
- 本文提出的PointLAM通过拉普拉斯点采样器和局部哈达玛聚合器来解决下采样和局部建模延迟问题。
- 实验结果表明,PointLAM在nuScenes和Waymo数据集上表现优异,尤其在小目标检测和处理稀疏数据方面具有明显优势。
📝 摘要(中文)
从LiDAR点云进行3D目标检测面临基本困境:体素方法在效率上有优势,但牺牲了几何精度;而点云方法则保持了高保真度,但计算瓶颈严重。为了解决这些系统性低效问题,本文提出了PointLAM,一种高效且强大的点云架构,结合了拉普拉斯点采样器(LPS)和局部哈达玛聚合器(LHA)两项创新。LPS通过隐式离散拉普拉斯高通滤波器和双重排序采样实现快速、结构感知的前景保留;LHA则通过瞬态网格解耦空间索引与特征表示,采用哈达玛门控机制替代复杂的连续交互。PointLAM在nuScenes和Waymo数据集上表现出色,展现了在小目标检测和极端稀疏情况下的显著优势。
🔬 方法详解
问题定义:本文旨在解决点云3D目标检测中的效率与精度矛盾,现有方法在保持高保真度的同时面临计算瓶颈,尤其是慢速下采样和昂贵的动态邻域查询。
核心思路:PointLAM通过引入拉普拉斯点采样器(LPS)和局部哈达玛聚合器(LHA)来优化下采样和局部建模过程,提升整体检测效率。LPS实现快速的结构感知前景保留,而LHA则通过瞬态网格和哈达玛门控机制简化了特征交互。
技术框架:PointLAM的整体架构包括LPS模块用于高效下采样,LHA模块用于局部特征聚合,结合双向Mamba(BDM)层进行全局序列建模,形成局部注意力Mamba(LAM)块。
关键创新:最重要的技术创新在于LPS和LHA的设计,LPS通过高通滤波和双重排序实现高效下采样,而LHA通过解耦空间索引与特征表示,采用哈达玛门控机制来提高局部建模的效率。
关键设计:在参数设置上,LPS和LHA的设计考虑了点云的稀疏性和结构特征,损失函数采用了针对目标检测的优化策略,网络结构则结合了局部与全局信息的有效融合。
🖼️ 关键图片
📊 实验亮点
在nuScenes和Waymo数据集上的实验结果显示,PointLAM在小目标检测和处理极端稀疏数据方面表现优异,性能与高度优化的体素方法相当,但计算开销仅为其一小部分,展现出显著的优势。
🎯 应用场景
该研究在自动驾驶、机器人导航和智能监控等领域具有广泛的应用潜力。通过提高点云3D目标检测的效率与精度,PointLAM能够为实时系统提供更可靠的环境感知能力,推动相关技术的发展与应用。
📄 摘要(原文)
3D object detection from LiDAR point clouds faces a fundamental dilemma: voxel-based methods achieve efficiency at the cost of geometric quantization, while point-based methods preserve fidelity but suffer from prohibitive computational bottlenecks. Specifically, point-based architectures are crippled by slow downsampling strategies (e.g., FPS) and expensive dynamic neighbor queries (e.g., k-NN) coupled with costly continuous interactions. To tackle these systemic inefficiencies, we propose PointLAM, a highly efficient and powerful point-based architecture driven by two synergistic innovations. First, to resolve the downsampling bottleneck, we develop the Laplacian Point Sampler (LPS). LPS employs an implicit discrete Laplacian high-pass filter and Doubly Sorted Sampling to achieve fast, structure-aware foreground preservation. Second, to overcome local modeling latency, we design the Local Hadamard Aggregator (LHA). LHA decouples spatial indexing from feature representation using transient grids, and replaces complex continuous interactions with a Hadamard Gating mechanism for topology-aware, attentive modulation. By coupling this local gating with Bi-Directional Mamba (BDM) layers for global sequence modeling, we formulate the Local Attentive Mamba (LAM) block. Powered by this architecture, PointLAM achieves competitive performance on nuScenes and Waymo for point-based detectors. It rivals highly optimized voxel competitors while requiring a fraction of the computational footprint, demonstrating marked superiority in detecting small instances and handling extreme sparsity. Project page: https://pointlam.github.io/.