{
  "paper_id": "0061e10a4aca3a8088ff50597f337d4b",
  "paper_title": "Automatic Labeled LiDAR Data Generation and Distance-Based Ensemble Learning for Human Segmentation",
  "paper_authors": [
    "WONJIK KIM",
    "MASAYUKI TANAKA",
    "MASATOSHI OKUTOMI",
    "YOKO SASAKI"
  ],
  "publication_date": "2019-05-06",
  "publication_venue": "IEEE Access",
  "abstract": "Following the improvements in deep neural networks, state-of-the-art networks have been proposed for human segmentation using point clouds captured by light detection and ranging. However, the performance of these networks depends significantly on the training data. An issue with collecting training data is labeling. Labeling by humans is necessary to obtain ground-truth labels; however, labeling involves high costs. Therefore, we propose an automatically labeled data generation pipeline, for which we can change any parameters or data generation environments. Our approach uses a human model named Dhaiba and the background of Miraikan to generate realistic artificial data. We present 1M data generated by the proposed pipeline. Furthermore, we propose an ensemble learning based on generated data for utilizing our data generation pipeline. This paper proposes the specifications of the pipeline, data details, and explanation of ensemble learning with evaluations of various approaches.",
  "keywords": [
    "Artificial intelligent",
    "machine learning",
    "computer vision",
    "LiDAR"
  ],
  "paper_institutions": [
    "Department of Systems and Control Engineering, School of Engineering, Tokyo Institute of Technology",
    "Artificial Intelligence Research Center, National Institute of Advanced Industrial Science and Technology"
  ],
  "output_language": "zh",
  "provider": "doubao",
  "model": "ark-code-latest",
  "prompt_template": "simple",
  "extracted_at": "2026-08-16T19:46:55.911214Z",
  "template_tag": "simple",
  "summary": "### 摘要\n本文针对基于LiDAR点云的人体分割中训练数据标注成本高的问题，提出了一种自动标注的数据生成流水线。该流水线使用Dhaiba人体模型和Miraikan背景生成逼真的合成数据，并自动生成标签，无需人工标注。作者生成了100万条数据，并提出了基于距离的集成学习方法，通过训练多个针对不同距离范围的网络并结合其输出，提高了分割性能。实验表明，该方法在生成数据和真实数据上均优于简单使用整体数据训练的网络。\n### 关键要点\n- 提出了一个三步数据生成流水线：背景数据收集、人体模型构建、LiDAR数据生成与自动标签生成。\n- 使用Dhaiba数字人体模型和Miraikan博物馆的室内背景，通过调节参数可生成任意距离、尺寸和类型的人体点云。\n- 生成了100万条标注数据，包含深度图、xyz坐标和人类标签，并以HDF5格式提供。\n- 设计了五种不同距离范围的数据集（1-10m、5-15m、10-20m、15-25m、20-30m），每个范围包含10K训练样本和0.1K测试样本。\n- 评估了FCN、FCDN和PointNet三种网络，其中FCN在深度输入下表现最佳，生成数据上的平均准确率为0.6895，真实数据上为0.6780。\n- 提出距离基高斯权重集成方法，利用输入深度图计算各网络权重，在生成数据上平均准确率提升至0.7813，真实数据上提升至0.6912。\n- 简单平均集成方法表现较差，因为各网络对未知距离范围的预测不可靠。\n- 训练权重集成方法在远距离数据集（15-25m、20-30m）上表现最佳，但整体不如高斯权重。\n- 增加训练数据量可提升平均准确率、召回率和威胁分数，但精度在0.1K时最高。\n- 该数据生成流水线可减少人工标注成本，适用于机器人导航等需要LiDAR人体分割的场景。",
  "available_templates": [
    "deep_read",
    "simple"
  ]
}