{
  "paper_id": "60620c70906458beb2d15e000c75c1b3",
  "paper_title": "OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields",
  "paper_authors": [
    "Zhe Cao",
    "Gines Hidalgo",
    "Tomas Simon",
    "Shih-En Wei",
    "Yaser Sheikh"
  ],
  "publication_date": "2019",
  "publication_venue": "IEEE Transactions on Pattern Analysis and Machine Intelligence",
  "abstract": "Realtime multi-person 2D pose estimation is a key component in enabling machines to have an understanding of people in images and videos. In this work, we present a realtime approach to detect the 2D pose of multiple people in an image. The proposed method uses a nonparametric representation, which we refer to as Part Affinity Fields (PAFs), to learn to associate body parts with individuals in the image. This bottom-up system achieves high accuracy and realtime performance, regardless of the number of people in the image. In previous work, PAFs and body part location estimation were refined simultaneously across training stages. We demonstrate that a PAF-only refinement rather than both PAF and body part location refinement results in a substantial increase in both runtime performance and accuracy. We also present the first combined body and foot keypoint detector, based on an internal annotated foot dataset that we have publicly released. We show that the combined detector not only reduces the inference time compared to running them sequentially, but also maintains the accuracy of each component individually. This work has culminated in the release of OpenPose, the first open-source realtime system for multi-person 2D pose detection, including body, foot, hand, and facial keypoints.",
  "keywords": [
    "2D human pose estimation",
    "2D foot keypoint estimation",
    "real-time",
    "multiple person",
    "part affinity fields"
  ],
  "paper_institutions": [
    "University of California, Berkeley",
    "Carnegie Mellon University",
    "Facebook Reality Labs"
  ],
  "output_language": "zh",
  "provider": "doubao",
  "model": "ark-code-latest",
  "prompt_template": "simple",
  "extracted_at": "2026-08-16T19:46:31.867322Z",
  "template_tag": "simple",
  "summary": "### 摘要\n\n本文针对多人物2D姿态估计这一关键问题，提出了基于部分亲和场（PAF）的实时自底向上方法。该方法利用非参数化的PAF表示来学习身体部位与个体的关联，避免了自顶向下方法对人物检测的依赖，并实现了推理时间与人数无关。相关工作包括基于图模型和深度学习的单人和多人姿态估计，但现有自底向上方法推理成本高。本文通过仅细化PAF分支而移除身体置信图细化，同时加深网络（将7x7卷积替换为3层3x3卷积），在MPII数据集上达到75.6%的mAP（3尺度搜索），在COCO验证集上达到65.3%的AP，速度约为22 FPS（GTX 1080 Ti）。此外，本文构建了包含约1.5万脚部关键点标注的脚部数据集，并训练了首个身体+脚部联合检测器，在保持精度的同时降低了推理时间。实验证明PAF细化比置信图细化更重要（仅PAF细化提高约7%精度和约2倍速度）。核心机制是PAF编码四肢的位置和方向，通过线积分度量候选部件间关联，并将多人体解析分解为多个二部图匹配的贪心推断。该工作已开源为OpenPose，支持身体、脚、手和面部关键点检测。未来可扩展至车辆关键点等任务。现实场景包括拥挤、遮挡、尺度变化等，但存在标注不全、混淆等约束。\n\n### 关键要点\n\n- 提出自底向上的多人体姿态估计方法，使用部分亲和场（PAF）表示部件间的关联，无需依赖人物检测器。\n- 通过仅细化PAF分支并移除身体置信图细化，网络深度增加而计算量减少，在MPII上精度提升约7%，速度提升约2倍。\n- 用3个3x3卷积替代1个7x7卷积，并拼接各层输出，增加非线性层次，同时降低运算量（从97降到51次操作）。\n- 在MPII测试集上（3尺度），达到75.6%的mAP（PCKh），相比DeeperCut提高超过13个百分点，推理时间仅为0.005秒/图。\n- 在COCO验证集上，新模型达到65.3%的AP（OKS指标），比原始版本提高6.9%，且无需CPM后处理。\n- 实验证明PAF细化比置信图细化更重要：仅用3个PAF阶段和1个置信图阶段（共4阶段）即优于先预测置信图的6阶段模型。\n- 构建了包含约1.5万个脚部关键点实例的脚部数据集，并公开，训练首个身体+脚部联合检测器，在保持身体精度时脚部AP达77.9%，且推理时间比分开运行快。\n- 在COCO上，自底向上方法的推理时间与人数无关，而自顶向下方法（Mask R-CNN, Alpha-Pose）随人数线性增加。\n- 贪婪二部图匹配的解析方法在精度上接近全局最优，但计算量极低，因为PAF的全局上下文由大感受野网络隐式编码。\n- 脚部标注考虑了深度，且身体+脚模型能改善踝关节等身体关键点估计。\n- 在车辆关键点任务上验证了方法通用性，无需修改架构即达到70.1% AP。\n- 常见的失败案例包括罕见姿态、遮挡、人群重叠、动物误检等，可通过更多数据增强或负样本缓解。"
}