{
  "paper_id": "cd46a6ff699b5f565d53b74f8712c99a",
  "paper_title": "A survey on deep 3D human pose estimation",
  "paper_authors": [
    "Rama Bastola Neupane",
    "Kan Li",
    "Tesfaye Fenta Boka"
  ],
  "publication_date": "2024-11-25",
  "publication_venue": "Artificial Intelligence Review",
  "abstract": "3D Human Pose Estimation (3D-HPE) is a highly active and evolving research area in computer vision with numerous applications such as extended reality, action recognition, and video surveillance. The field has significantly advanced with deep learning, public datasets, and enhanced computational power, addressing challenges like depth ambiguity, occlusion, and data scarcity. Researchers confront scenario-specific issues such as ill-posed problems in monocular setups, cross-view aggregation with camera synchronizations in multi-view systems, and inter-person occlusion in multi-person scenarios. This survey comprehensively reviews contemporary strategies covering a technological spectrum including Convolutional Neural Networks, Graph Convolutional Networks, Transformers, and their combinations employed to address these challenges. It includes scenarios such as monocular and multi-view setups, single and multi-person cases, as well as image and video inputs. The survey explores various solution paradigms, including single-stage vs 2D-to-3D lifting, absolute vs relative keypoints, pixel vs voxel vs Neural Radiance Field spaces, and deterministic, probabilistic, or diffusion-based strategies, along with top-down vs bottom-up approaches. It examines advanced learning techniques beyond supervised methods and data augmentation for diverse pose datasets. It analyzes the performance of recent methods on benchmark datasets for different scenarios. Challenges are categorized into common and scenario-specific issues, and future research directions are proposed to foster further advancements in the field. Additionally, key sections are summarized in tables or visual formats for quick understanding. This survey is a valuable resource and a solid reference for researchers in the dynamic landscape of 3D human pose estimation.",
  "keywords": [
    "3D human pose estimation",
    "Single-person",
    "Multi-person",
    "Data augmentation",
    "Learning techniques",
    "Scenario-specific challenges"
  ],
  "paper_institutions": [],
  "output_language": "zh",
  "provider": "doubao",
  "model": "ark-code-latest",
  "prompt_template": "simple",
  "extracted_at": "2026-08-16T19:47:14.548384Z",
  "template_tag": "simple",
  "summary": "### 摘要\n本综述系统梳理了深度三维人体姿态估计（3D-HPE）领域的最新进展。针对单目/多视角、单人与多人、图像与视频等八种场景，重点分析了CNN、GCN、Transformer及其混合架构的解决方案。文章探讨了单阶段与2D-to-3D提升、绝对与相对关键点坐标、像素/体素/NeRF求解空间、确定性/概率/扩散模型、自顶向下与自底向上等范式。为应对数据稀缺与多样性不足，综述总结了半监督、弱监督、自监督及在线/离线姿态数据增强技术。在数据集方面，本文比较了Human3.6M、MPI-INF-3DHP、MuPoTS-3D等基准上的性能，并分析了常用评估指标。综述将挑战归纳为共性（如深度歧义、遮挡、标注不足）与场景特有（如跨视角聚合、多人交互）两类，并提出了包括基于LLM、实时轻量化、动态场景交互在内的未来研究方向，旨在为研究者提供全面参考。核心机制方面，2D-to-3D提升方法通过先预测2D关键点再利用GCN/Transformer等建模关节关系统，可有效降低深度歧义并利用丰富2D数据。\n\n### 关键要点\n- 3D-HPE根据相机数量与人数可分为八种场景，每种场景面临不同挑战，例如单目存在深度歧义，多视角需要相机标定与同步。\n- 单阶段端到端方法直接回归3D坐标，但计算量大且依赖3D标注；2D-to-3D提升方法利用成熟的2D姿态检测器，可减少过拟合且适应野外数据。\n- 关键点坐标可分为人物中心（person-centric）和相机中心（camera-centric），多人场景通常需要绝对坐标以反映真实位置。\n- 多人姿态估计采用自顶向下、自底向上或混合范式，自顶向上下精度高但随人数增加计算量增大，自底向上效率高但对尺度变化敏感。\n- 概率与扩散方法可为每个输入生成多个假设，捕捉深度歧义带来的不确定性，但需要聚合策略得到最终结果。\n- 求解空间包括像素、体素与NeRF，体素方法利于多视角融合但受内存和分辨率限制，NeRF可合成新视图辅助遮挡处理。\n- 半监督、自监督等学习技术利用未标注数据或几何一致性约束，缓解3D标注稀缺问题。\n- 姿态数据增强分为离线与在线两类，离线方法生成多样化2D-3D对，在线方法通过可微分增强器与估计器协同训练。\n- 常用数据集如Human3.6M、MPI-INF-3DHP、CMU Panoptic等各有侧重，评估指标包括MPJPE、PA-MPJPE、PCK等。\n- 性能比较显示，单目单人的图像任务中GCN方法表现优越，视频任务中GCN+Transformer混合方法更优；多视角多人任务中Transformer和体素CNN方法性能领先。\n- 未来方向包括利用大语言模型提升语义理解、设计轻量模型用于实时应用、结合场景感知处理动态交互。\n- 挑战分类包括共性挑战（人体复杂结构、深度歧义、标注稀缺等）和场景特有挑战（单目泛化、多视角标定、多人遮挡等）。"
}