{
  "paper_id": "c73a546869cf17e86b38a831963a3822",
  "paper_title": "Privacy-preserving People Detection Enabled by Solid State LiDAR",
  "paper_authors": [
    "Andrei Günter",
    "Stephan Böker",
    "Matthias König",
    "Martin Hoffmann"
  ],
  "publication_date": "",
  "publication_venue": "2020 16th International Conference on Intelligent Environments (IE)",
  "abstract": "Detecting people in video streams involves privacy concerns since people are often unaware of being recorded. Furthermore, cameras and commodity depth sensors are prone to variances of illumination. This work proposes light detection and ranging as sensing technology for a robust and privacy preserving people detection. A people counter is implemented, which captures a coarse human shape by concatenating multiple two dimensional range scans while people pass the sensor's view. The system works under dynamic illuminated conditions and is running as an embedded node on a single core 700 MHz processing unit. An evaluation with 20 different people crossing an entrance 100 times in each direction was carried out under controlled conditions. Results show a high precision and recall of 100 percent and 99 percent respectively for counting people and determining the walking direction. Privacy is preserved since objects are represented by distance values with a low resolution of 16x16 data points.",
  "keywords": [
    "LiDAR",
    "people detection",
    "privacy preservation",
    "people counting",
    "solid-state LiDAR"
  ],
  "paper_institutions": [
    "Bielefeld University of Applied Sciences"
  ],
  "output_language": "zh",
  "provider": "doubao",
  "model": "ark-code-latest",
  "prompt_template": "simple",
  "extracted_at": "2026-08-16T19:46:38.995594Z",
  "template_tag": "simple",
  "summary": "### 摘要\n本工作针对视频流中人员检测的隐私问题以及相机和传统深度传感器受光照变化影响的问题，提出使用固态激光雷达（LiDAR）作为传感技术，实现鲁棒且保护隐私的人员检测。系统通过拼接多个二维距离扫描捕获行人的粗略人体轮廓，用于实现人员计数器。系统在动态光照条件下运行，并部署在单核700 MHz处理单元的嵌入式节点上。在受控条件下，20名不同人员各100次通过入口进行测试，结果显示人员计数和行走方向判定的精确率与召回率分别达到100%和99%。隐私通过低分辨率（16×16数据点）的距离值表示得以保护。相关工作包括基于高分辨率深度数据的方法，但其存在隐私泄露风险；本工作采用低分辨率传感器来确保隐私。方案中，传感器倾斜安装以区分行走方向，数据预处理包括分割和归一化，分类采用两个SVM进行二分类。核心机制为关键帧提取，通过计算相邻帧差异δ_k = ∑(y_{x,k+1} - y_{x,k})²来筛选冗余帧，最终选取16个关键帧表示对象。实验基于2000个样本（1000个人类，1000个非人类）训练，采用10折交叉验证。未来的工作包括改进数据集以覆盖更多姿态和物体，以及在真实环境中进行更全面的评估。\n\n### 关键要点\n- 采用固态激光雷达（Leddar M16）进行人员检测，不受光照变化影响，且具有隐私保护优势。\n- 系统以16×16的数据点表示人体，低分辨率确保无法识别个体身份。\n- 传感器安装在入口上方，倾斜放置以区分行走方向（进入或离开）。\n- 预处理包括对象分割和归一化，帧序列被压缩为固定数量的关键帧（每对象16帧）。\n- 关键帧提取使用差异阈值，基于相邻帧的平方差之和来判断冗余帧。\n- 分类采用两个SVM：第一个区分人类与非人类，第二个区分行走方向。\n- 训练数据集包含2000个样本，其中1000个为人类（高度165-190cm），1000个为非人类（动物、推车、随机数据）。\n- 使用10折交叉验证调整SVM参数，多项式核维度和软间隔常数分别设定（人类/非人类：n=5, C=0.025；方向：n=2, C=0.1）。\n- 评估中20人（13人未参与训练）各通过入口5次进入和5次离开，共200次穿越。\n- 结果显示人类检测的精确率为100%，召回率为99.5%；进入和离开的精确率均为100%，召回率分别为100%和99%。\n- 系统功耗仅为4.7W，适合嵌入式部署，适用于购物中心等频繁访客场所。\n- 局限性：当前数据集未包含戴帽子、携带物品或坐轮椅的人，可能影响检测准确性，需扩展数据集和预处理方法。"
}