{
  "paper_id": "bf3dd1eda58da2c55660ad61138b38af",
  "paper_title": "Efficient Online Transfer Learning for Road Participants Detection in Autonomous Driving",
  "paper_authors": [
    "Rui Yang",
    "Zhi Yan",
    "Tao Yang",
    "Yaonan Wang",
    "Yassine Ruichek"
  ],
  "publication_date": "2023-08-18",
  "publication_venue": "IEEE Sensors Journal",
  "abstract": "The spatial information provided by 3-D object detection is critical for autonomous driving. Although visual-based object detection has made significant progress in 2-D/3-D in recent years, nonvisual sensors such as 3-D light detection and ranging (LiDAR) still have inherent advantages in the accuracy (ACC) of object localization. However, the challenge still lies in the interpretability of the sparse point clouds it generates and the difficulty of being manually annotated. In this article, we propose an online transfer learning framework based on multimodal sensor systems for 3-D object detection of urban road participants including pedestrians, cyclists, and cars. The framework aims to automatically and efficiently transfer object detection capabilities from 2-D monocular camera to 3-D LiDAR through a multitarget tracker-based pipeline, enabling knowledge transfer between sensors of different modalities. Furthermore, online random forest (ORF), an inherently fast and multiclass learning method, is innovatively integrated into our system. Experiments on two very different datasets, including KITTI and Waymo, demonstrate that the proposed framework not only rapidly builds 3-D detection capabilities about road participants in a single dataset but also maintains this capability across datasets. This reveals that the proposed framework is particularly suitable for in situ deployment of unmanned vehicles and can also solve the problem of insufficient generalization ability of current offline-trained detectors.",
  "keywords": [
    "3-D object detection",
    "autonomous driving",
    "light detection and ranging (LiDAR)",
    "multimodal perception",
    "online machine learning"
  ],
  "paper_institutions": [
    "CIAD UMR 7533, Univ. Bourgogne Franche-Comté, UTBM, F-25200 Montbéliard, France",
    "Unmanned System Research Institute, Northwestern Polytechnical University, Xi'an 710072, China",
    "College of Electrical and Information Engineering, Hunan University, Changsha 410082, China",
    "National Engineering Research Center for Robot Visual Perception and Control Technology, Changsha 410082, China"
  ],
  "output_language": "zh",
  "provider": "doubao",
  "model": "ark-code-latest",
  "prompt_template": "simple",
  "extracted_at": "2026-08-16T19:47:37.308904Z",
  "template_tag": "simple",
  "summary": "### 摘要\n本论文针对自动驾驶中基于3D LiDAR的目标检测面临的数据标注困难、环境适应能力差等问题，提出了一种基于多模态传感器系统的在线迁移学习框架。该框架利用预训练的2D单目相机检测器作为知识源，通过多目标跟踪器构建从2D图像到3D点云的知识迁移管道，实现无需人工标注的3D道路参与者（行人、骑车人、汽车）检测能力在线学习。核心创新在于将在线随机森林（ORF）集成到系统中，支持快速多分类学习。在KITTI和Waymo两个差异显著的数据集上进行实验，结果表明所提框架不仅能快速在单一数据集上建立3D检测能力，还能跨数据集保持该能力，特别适合无人车的原位部署，并解决了当前离线训练检测器泛化能力不足的问题。实验采用平均精度（AP）和宏平均（MaA）等指标，验证了多模态融合检测的优越性和在KITTI基准上的竞争力。\n\n### 关键要点\n- 提出一个在线迁移学习框架，利用2D相机检测器辅助3D LiDAR学习目标检测，实现跨模态知识迁移。\n- 框架包含四个模块：现成检测器、学习即用检测器、多目标跟踪器和样本标注器，通过闭环迭代实现“边用边学”。\n- 采用在线随机森林（ORF）作为分类器，支持流式数据、小批量学习和实时模型存储，并满足嵌入式部署需求。\n- 使用地面平面拟合（GPF）进行地面滤波，Autoware聚类进行点云分割，提取61维特征（包括点数、距离、协方差等）用于分类。\n- 多目标跟踪器结合PDA、UKF和IMM，实现检测结果的时空关联，构建物体轨迹。\n- 样本标注器基于轨迹概率融合多检测器置信度，对轨迹样本进行自动标注，生成训练数据。\n- 在KITTI数据集上，Integrated-ORF与Standalone-ORF相比，在汽车类别上略有优势，但在行人和骑车人上表现稍逊，表明无监督学习仍有提升空间。\n- 在KITTI 3D检测验证集上，融合现成检测器的多模态方法显著提升了AP值，如汽车easy难度从65.12%提高到80.45%。\n- 在KITTI 2D检测验证集上，添加学习即用检测器后，2D检测性能也得到提升，如汽车moderate难度从78.18%提高到90.45%。\n- 在KITTI测试集上与MV3D、AVOD、F-Pointnet等离线方法比较，所提方法在骑车人类别上达到最优，且总体表现接近最佳。\n- 跨数据集实验（从KITTI到Waymo）显示，初始性能下降但随在线学习迭代逐渐恢复，适应新环境。\n- 消融实验表明，引入空间关联、时间关联和闭环学习逐步提升分类性能，验证各模块贡献。\n- 定性分析显示，点云与图像匹配良好，但存在欠分割问题（如骑车人与电线杆粘连），可通过引入颜色纹理信息进一步改进。"
}