{
  "paper_id": "8101d07ebc10e8f6b482e489968fc48e",
  "paper_title": "Learning-Based Human Segmentation and Velocity Estimation Using Automatic Labeled LiDAR Sequence for Training",
  "paper_authors": [
    "WONJIK KIM",
    "MASAYUKI TANAKA",
    "MASATOSHI OKUTOMI",
    "YOKO SASAKI"
  ],
  "publication_date": "2020-05-08",
  "publication_venue": "IEEE Access",
  "abstract": "In this paper, we propose an automatic labeled sequential data generation pipeline for human segmentation and velocity estimation with point clouds. Considering the impact of deep neural networks, state-of-the-art network architectures have been proposed for human recognition using point clouds captured by Light Detection and Ranging (LiDAR). However, almost all conventional datasets are either a collection of single LiDAR scanning with label information or sequential LiDAR scanning without label information. This limitation has disturbed the progress of research to date. Therefore, we have developed an automatic labeled sequential data generation pipeline, in which we can control any parameter or data generation environment with pixel-wise and per-frame ground truth segmentation and pixel-wise velocity information for human recognition. Our approach uses a precise human model and reproduces a precise motion to generate realistic artificial data. We present more than 7K sequences, where each sequence consists of 32 frames generated by the proposed pipeline. With the proposed sequence generator, we confirm that human segmentation performance is improved when using the sequential data compared to when using the data from a single LiDAR scan. We also evaluate our data by comparing with data generated under different conditions. In addition, we estimate pedestrian velocity with LiDAR by only utilizing data generated by the proposed pipeline.",
  "keywords": [
    "Artificial intelligent",
    "machine learning",
    "computer vision",
    "LiDAR"
  ],
  "paper_institutions": [
    "Tokyo Institute of Technology",
    "National Institute of Advanced Industrial Science and Technology"
  ],
  "output_language": "zh",
  "provider": "doubao",
  "model": "ark-code-latest",
  "prompt_template": "simple",
  "extracted_at": "2026-08-16T19:46:53.321674Z",
  "template_tag": "simple",
  "summary": "### 摘要\n本论文针对LiDAR点云中的行人分割与速度估计问题，提出了一种全自动标注序列数据生成流程。现有数据集要么是带标签的单帧扫描，要么是无标签的序列扫描，限制了基于深度学习的方法发展。该方法利用精确的人体模型（基于DhaibaWorks）和步行运动模型，结合人体轨迹与LiDAR轨迹，生成具有像素级标签和速度信息的序列数据。生成的数据包含超过7K个序列，每个序列32帧，并提供了32×1024的深度图、xyz坐标、人类标签及速度图。实验采用基于FCN扩展的多帧输入网络架构，通过共享卷积层提取特征后拼接，并同时输出分割与速度估计。在评估中，使用IoU指标衡量分割性能，使用MSE衡量速度估计。结果表明，序列数据相比单帧数据在所有距离范围内都提升了IoU，尤其在远距离（8m以上）提升显著，例如16帧在真实数据上IoU为0.5883，而单帧为0.5097。消融实验显示，加入轨迹和步行模型显著提高了性能（IoU从0.2276升至0.3814）。速度估计任务中，估计速度与真实速度趋势一致，证实了仅用LiDAR进行速度估计的可行性。该流程还支持任意参数控制，且数据已公开。未来工作包括扩展人体模型多样性（如姿势、服装）以及应用于跟踪和轨迹预测等任务。核心机制是通过像素级速度公式 $\\boldsymbol{v}_{p}=\\begin{cases}\\boldsymbol{V}_{h}-\\boldsymbol{V}_{s}&if~human~pixel\\\\-\\boldsymbol{V}_{s}&if~background~pixel\\end{cases}$ 合成速度信息。\n\n### 关键要点\n- 提出了全自动生成带标签LiDAR序列的流程，无需人工标注，能够控制任何生成参数。\n- 使用DhaibaWorks构建精确的人体模型，结合观察到的步行运动数据和实际采集的行人轨迹生成多样化场景。\n- 生成了超过7K个序列，每个序列32帧，包含像素级分割标签和速度信息，数据已公开。\n- 网络架构基于FCN扩展，支持多帧输入，通过共享权重卷积提取特征然后拼接，同时进行分割和速度估计。\n- 实验对比了单帧和不同帧数（2、4、8、16）的序列，序列方法在所有距离范围内均优于单帧，尤其是远距离提升明显。\n- 在真实数据上，16帧网络的IoU为0.5883，相比单帧的0.5097提升了约15.4%。\n- 消融实验表明，轨迹和步行模型对性能有积极影响，完整数据训练的网络IoU最高（0.3814）。\n- 增加速度估计任务后，分割性能进一步提升（生成数据IoU从0.5709升至0.6443），表明速度信息有益于分割。\n- 速度估计结果与真实值趋势一致，验证了仅用LiDAR进行速度估计的可行性。\n- 数据集生成参数包括32线LiDAR、1024水平分辨率、10Hz采样率等，模拟真实传感器配置。\n- 当前网络仅用生成数据训练，真实数据仅用于评估，未来可探索生成数据与真实数据混合训练。\n- 未来研究方向包括改进人体模型的真实度（如姿势、服装），并扩展到跟踪、轨迹预测等应用。"
}