{
  "paper_id": "8f1e66ac927cb35bfe6bb39257e9ceb1",
  "paper_title": "Neighborhood-Enhanced 3D Human Pose Estimation with Monocular LiDAR in Long-Range Outdoor Scenes",
  "paper_authors": [
    "Jingyi Zhang",
    "Qihong Mao",
    "Guosheng Hu",
    "Siqi Shen",
    "Cheng Wang"
  ],
  "publication_date": "",
  "publication_venue": "Proceedings of the AAAI Conference on Artificial Intelligence",
  "abstract": "3D human pose estimation (3HPE) in large-scale outdoor scenes using commercial LiDAR has attracted significant attention due to its potential for real-life applications. However, existing LiDAR-based methods for 3HPE primarily rely on recovering 3D human poses from individual point clouds, and the coherence cues present in the neighborhood are not sufficiently harnessed. In this work, we explore spatial and contextual coherence cues contained in the neighborhood that leads to great performance improvements in 3HPE. Specifically, firstly, we deeply investigate the 3D neighbor in the background (3BN) which serves as a spatial coherence cue for inferring reliable motion since it provides physical laws to limit motion targets. Secondly, we introduce a novel 3D scanning neighbor (3SN) generated during the data collection and 3SN implies structural edge coherence cues. We use 3SN to overcome the degradation of performance and data quality caused by the sparsity-varying properties of LiDAR point clouds. In order to effectively model the complementation between these distinct cues and build consistent temporal relationships across human motions, we propose a new transformer-based module called the CoherenceFuse module. Extensive experiments conducted on publicly available datasets, namely LidarHuman26M, CIMI4D, SLOPER4D and Waymo Open Dataset v2.0, showcase the superiority and effectiveness of our proposed method. In particular, when compared with LidarCap on the LidarHuman26M dataset, our method demonstrates a reduction of 7.08mm in the average MPJPE metric, along with a decrease of 16.55mm in the MPJPE metric for distances exceeding 25 meters. The code and models are available at https://github.com/jingyi-zhang/Neighborhoodenhanced-LidarCap.",
  "keywords": [
    "3D human pose estimation",
    "LiDAR",
    "long-range",
    "neighborhood coherence",
    "transformer",
    "CoherenceFuse"
  ],
  "paper_institutions": [
    "Fujian Key Laboratory of Sensing and Computing for Smart Cities, Xiamen University, China",
    "Key Laboratory of Multimedia Trusted Perception and Efficient Computing, Ministry of Education of China, School of Informatics, Xiamen University, China"
  ],
  "output_language": "zh",
  "provider": "doubao",
  "model": "ark-code-latest",
  "prompt_template": "simple",
  "extracted_at": "2026-08-16T19:47:39.519002Z",
  "template_tag": "simple",
  "summary": "### 摘要\n本文针对长距离户外场景中基于单目LiDAR的3D人体姿态估计（3HPE）问题。现有的LiDAR方法主要从单人点云中恢复姿态，忽略了邻域中的一致性线索。本文提出利用两类邻域一致性线索：3D背景邻居（3BN）提供空间一致性线索，通过物理定律约束运动目标；3D扫描邻居（3SN）由LiDAR扫描特性产生，携带结构边缘一致性线索，可补偿远距离点云稀疏性导致的质量下降。为融合这些互补线索并建模时间关系，提出了基于Transformer的CoherenceFuse模块。实验在LidarHuman26M、CIMI4D、SLOPER4D和Waymo Open Dataset v2.0四个公开数据集上进行，结果表明所提方法显著优于现有方法。在LidarHuman26M上与LidarCap相比，MPJPE平均降低7.08mm，距离超过25米时降低16.55mm。\n\n### 关键要点\n- 提出了3D背景邻居（3BN）概念，利用其空间一致性线索为3HPE提供物理先验，增强运动估计的可靠性。\n- 提出了3D扫描邻居（3SN），利用激光束在人体边缘继续传播的特性，获取结构边缘信息，缓解远距离点云退化。\n- 设计了CoherenceFuse模块，包含自注意力和交叉注意力，有效融合3BN、3SN和人体点云特征，并建模时间关系。\n- 在LidarHuman26M数据集上相比LidarCap，MPJPE降低7.08mm，PCK30提升1.94%。\n- 在距离超过25米的情况下，MPJPE降低16.55mm，PCK30提升5.64%，验证了3SN对长距离场景的有效性。\n- 在CIMI4D（严重自遮挡）上，MPJPE比LidarCap降低8.66mm；在SLOPER4D（复杂现实环境）上降低5.9mm。\n- 在Waymo Open Dataset v2.0上，CD指标比LidarCap降低4.79mm，比P4T降低7.02mm，证明泛化能力。\n- 消融实验表明，3BN和3SN单独都能提升性能，而CoherenceFuse模块融合效果最佳。\n- 3BN半径参数实验显示2米半径最优，过大背景信息反而无益。\n- 通过阴影点对比实验，证明3SN编码的不只是运动轮廓，而是LiDAR特有的结构边缘信息，远超简单投影。\n- 方法无需额外模态数据或物理约束项，运算高效，更适用于实时应用。\n- 未来工作将集成二值分割网络，处理整个数据帧中的前景分割和检测任务。"
}