{
 "id": "ep9hjeh2tv",
 "slug": "alphapose",
 "title": "AlphaPose",
 "updated": "2026-10-25",
 "topic_path": [
  {
   "id": "technology",
   "label": "Technology and the built world",
   "api_url": "https://www.edgepedia.org/api/v1/topics/technology"
  },
  {
   "id": "technology.computing",
   "label": "Computing and digital systems",
   "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing"
  },
  {
   "id": "technology.computing.ai",
   "label": "Artificial intelligence and data",
   "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai"
  },
  {
   "id": "technology.computing.ai.perception_language",
   "label": "Language and vision AI",
   "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai.perception_language"
  },
  {
   "id": "technology.computing.ai.computer_vision",
   "label": "Computer vision",
   "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai.computer_vision"
  },
  {
   "id": "technology.computing.ai.cv_methods",
   "label": "Vision methods and geometry",
   "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai.cv_methods"
  },
  {
   "id": "technology.computing.ai.cv_methods.pose_estimation",
   "label": "Pose estimation and tracking of pose",
   "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai.cv_methods.pose_estimation"
  }
 ],
 "geo": [
  {
   "id": "geo.nongeo.t2001.technology.computing.ai.perception_language",
   "label": "Non-geographic · 2001 to 2020: Language and vision AI",
   "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001.technology.computing.ai.perception_language",
   "path": [
    {
     "id": "geo.nongeo",
     "label": "Non-geographic",
     "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo"
    },
    {
     "id": "geo.nongeo.t2001",
     "label": "Non-geographic · 2001 to 2020",
     "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001"
    },
    {
     "id": "geo.nongeo.t2001.technology",
     "label": "Technology and the built world",
     "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001.technology"
    },
    {
     "id": "geo.nongeo.t2001.technology.computing",
     "label": "Computing and digital systems",
     "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001.technology.computing"
    },
    {
     "id": "geo.nongeo.t2001.technology.computing.ai",
     "label": "Artificial intelligence and data",
     "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001.technology.computing.ai"
    },
    {
     "id": "geo.nongeo.t2001.technology.computing.ai.perception_language",
     "label": "Language and vision AI",
     "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001.technology.computing.ai.perception_language"
    }
   ]
  }
 ],
 "excerpt": "AlphaPose is an open-source deep learning system for multi-person human pose estimation that detects people in images and video and returns their body keypoints as skeletons.",
 "snippet": "AlphaPose is an open-source deep learning system for multi-person human pose estimation that detects people in images and video and returns their body keypoints as skeletons.",
 "node": "technology.computing.ai.cv_methods.pose_estimation",
 "markdown": "# AlphaPose\n\nAlphaPose is an open-source deep learning system for multi-person human pose estimation that detects people in an image or video and returns their body keypoints as skeletons. It grew out of a framework called RMPE (Regional Multi-Person Pose Estimation) and was, according to its maintainers, the first open-source system to exceed 70 mAP on the COCO keypoints benchmark and 80 average PCKh (82.1 average PCKh) on MPII.<sup>[1](https://github.com/MVIG-SJTU/AlphaPose)</sup> A later whole-body version, which adds face, hand, and foot keypoints and joint tracking, was published in [IEEE Transactions on Pattern Analysis and Machine Intelligence](https://www.edgepedia.org/ieee-transactions-on-pattern-analysis-and-machine-intelligence) by Hao-Shu Fang and colleagues.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup>\n\n| Key fact | Value |\n|---|---|\n| Task | Multi-person 2D pose estimation (top-down: detect, then estimate pose per box)<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup> |\n| COCO keypoints accuracy | 73.3 AP@0.5:0.95 (AP@0.5 89.2, AP@0.75 79.1)<sup>[1](https://github.com/MVIG-SJTU/AlphaPose)</sup> |\n| MPII accuracy | 82.1 average PCKh (RMPE paper reports 76.7 mAP on MPII multi-person)<sup>[1](https://github.com/MVIG-SJTU/AlphaPose)</sup><sup> • </sup><sup>[3](https://ar5iv.labs.arxiv.org/html/1612.00137)</sup> |\n| Speed | Over 10 fps for pose estimation and tracking on a 2080Ti; Simple Baseline ResNet50 reaches 2.94 iter/s at 70.6 AP on a TITAN XP<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup><sup> • </sup><sup>[4](https://github.com/MVIG-SJTU/AlphaPose/blob/c60106d1/docs/MODEL_ZOO.md)</sup> |\n| Whole-body keypoints | 136 (body, face, hands, feet), 68 (no face), or 21 (single hand)<sup>[4](https://github.com/MVIG-SJTU/AlphaPose/blob/c60106d1/docs/MODEL_ZOO.md)</sup> |\n| Output format | COCO format by default, compatible with OpenPose<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup> |\n| Tracking | PoseFlow tracker: 66.5 mAP and 58.3 MOTA on PoseTrack Challenge<sup>[1](https://github.com/MVIG-SJTU/AlphaPose)</sup> |\n\n## How it works\n\nAlphaPose follows a top-down framework: it first detects human bounding boxes, then estimates the pose within each box independently.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup> This design inherits a weakness of single-person pose estimators, which fail when given inaccurate bounding boxes or redundant detections.<sup>[3](https://ar5iv.labs.arxiv.org/html/1612.00137)</sup> The system compensates in two ways. First, it lowers the detection confidence and NMS thresholds to provide more candidate boxes for pose estimation; the redundant poses produced from redundant boxes are then eliminated by a parametric pose NMS, which introduces a novel pose distance metric to compare pose similarity.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup> The parameters of this pose distance are optimized with a data-driven approach.<sup>[3](https://ar5iv.labs.arxiv.org/html/1612.00137)</sup>\n\nThe original RMPE framework consisted of three components: a Symmetric Spatial Transformer Network (SSTN), Parametric Pose Non-Maximum-Suppression, and a Pose-Guided Proposals Generator (PGPG).<sup>[3](https://ar5iv.labs.arxiv.org/html/1612.00137)</sup> The whole-body version adds Symmetric Integral Keypoint Regression (SIKR) for fast and fine localization, Parametric Pose NMS (P-NMS), and Pose Aware Identity Embedding for jointly performing pose estimation and tracking.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup>\n\n## How it is done\n\nIn practice, a user supplies an image directory or video and a trained detector and pose model. The detector (for example YOLOv3 or YOLOX) proposes person boxes; the pose network processes each cropped box; parametric pose NMS removes duplicates; and results are written as keypoint coordinates in COCO format.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup><sup> • </sup><sup>[1](https://github.com/MVIG-SJTU/AlphaPose)</sup> The whole-body models additionally regress face, hand, and foot keypoints in the same pass.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup>\n\nIn the current repository, inference is run via `./scripts/inference.sh {CONFIG} {CHECKPOINT} {VIDEO_NAME}` with an optional output directory, or through `demo_inference.py` with an optional yolox-x detector; the older PyTorch branch provides `demo.py` for image directories (`python3 demo.py --indir {img_directory} --outdir examples/res`) and `video_demo.py` for videos with `--save_video`.<sup>[1](https://github.com/MVIG-SJTU/AlphaPose)</sup> The default saving format is COCO format and can be made compatible with [OpenPose](https://www.edgepedia.org/openpose).<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup> AlphaPose is developed on both PyTorch and MXNet and supports both Linux and Windows.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup>\n\n## Origin\n\nAlphaPose is based on RMPE, a regional multi-person pose estimation framework.<sup>[3](https://ar5iv.labs.arxiv.org/html/1612.00137)</sup><sup> • </sup><sup>[1](https://github.com/MVIG-SJTU/AlphaPose)</sup> The whole-body system was reported by Hao-Shu Fang and colleagues in IEEE Transactions on Pattern Analysis and Machine Intelligence in 2022.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup> The repository lists the current maintainers as Jiefeng Li, Hao-shu Fang, Haoyi Zhu, Yuliang Xiu, and Chao Xu.<sup>[1](https://github.com/MVIG-SJTU/AlphaPose)</sup>\n\n## Variants\n\nThe repository supports YOLOX, YOLOV3-SPP, EfficientDet, and JDE detectors, and SimplePose, HRNet, and FastPose pose estimators, including FastPose-DCN variants, plus PoseFlow and Re-ID based tracking.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup> PoseFlow is described as the first open-source online pose tracker achieving both 60+ mAP (66.5 mAP) and 50+ MOTA (58.3 MOTA) on the PoseTrack Challenge dataset.<sup>[1](https://github.com/MVIG-SJTU/AlphaPose)</sup>\n\nSpeed and accuracy depend strongly on the backbone and input size. On COCO val2017, Fast Pose (DUC) ResNet152 with YOLOv3 at 256x192 input reaches 73.3 AP at 1.62 iter/s on a TITAN XP with batch_size=64, while Simple Baseline ResNet50 reaches 70.6 AP at 2.94 iter/s and Fast Pose (DCN) ResNet50 reaches 72.8 AP at 2.94 iter/s.<sup>[4](https://github.com/MVIG-SJTU/AlphaPose/blob/c60106d1/docs/MODEL_ZOO.md)</sup> The whole-body variants estimate face, body, hand, and foot keypoints jointly and come in 136-keypoint, 68-keypoint (no face), and 21-keypoint (single hand) forms.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup><sup> • </sup><sup>[4](https://github.com/MVIG-SJTU/AlphaPose/blob/c60106d1/docs/MODEL_ZOO.md)</sup> On COCO WholeBody (133 keypoints), Fast Pose (DCN) with combined loss reaches 58.2 AP at 10.22 iter/s, and Fast Pose (DUC) ResNet152 reaches 56.9 AP at 15.72 iter/s.<sup>[4](https://github.com/MVIG-SJTU/AlphaPose/blob/c60106d1/docs/MODEL_ZOO.md)</sup> The parallelized pipeline runs at over 10 fps for both pose estimation and tracking on a standard GPU such as a 2080Ti.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup>\n\n## Applications\n\nIn an independent benchmark of 16 frameworks on exercise videos, AlphaPose was among the frameworks with a fully functional demo that saved keypoint coordinates and generated skeleton-overlay videos, and it was one of the top 2D performers alongside rtmlib and YOLOv7.<sup>[5](https://ieeexplore.ieee.org/document/11429610)</sup> Published sources do not document concrete application deployments, for example in surveillance or sports analytics.\n\n## Limitations and alternatives\n\nAs a top-down method, AlphaPose depends on its first-stage detector. A systematic survey notes that because of missing detections and obscure bounding boxes in the first stage, the human postures estimated in top-down approaches are easily lost or misjudged.<sup>[6](https://link.springer.com/article/10.1007/s10462-024-11060-2)</sup> AlphaPose's own design mitigates this with lowered thresholds and parametric pose NMS rather than eliminating it.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup>\n\nAgainst earlier open-source systems, the repository's benchmark table gives AlphaPose 73.3 AP@0.5:0.95 on COCO keypoints versus 61.8 for OpenPose (CMU-Pose) and 67.0 for Detectron ([Mask R-CNN](https://www.edgepedia.org/mask-r-cnn)), and 82.1 versus 75.6 average PCKh on MPII.<sup>[1](https://github.com/MVIG-SJTU/AlphaPose)</sup> The maintainers also state that, benchmarked on a single Nvidia 2080Ti, AlphaPose is more efficient than OpenPose when there are fewer than 20 persons in the scene.<sup>[2](https://doi.org/10.1109/tpami.2022.3222784)</sup> Note that the repository's headline claim of 75 mAP on COCO and its benchmark table value of 73.3 AP differ; both figures come from the same official source and the discrepancy is not resolved in the official documentation.<sup>[1](https://github.com/MVIG-SJTU/AlphaPose)</sup>\n\nIn the IEEE benchmark of 16 frameworks on exercise videos, MeTRAbs emerged as the best overall framework; processing one hour of video ranged from 2 to 26 hours across frameworks, and runtime did not correlate with accuracy.<sup>[5](https://ieeexplore.ieee.org/document/11429610)</sup> RTMPose and rtmlib are documented as newer alternatives in the benchmark literature.<sup>[5](https://ieeexplore.ieee.org/document/11429610)</sup> Head-to-head numbers against HRNet-based methods and YOLO-pose pipelines beyond this benchmark have not been settled by published comparisons.\n\n## References\n\n1. [MVIG-SJTU/AlphaPose (official repository README)](https://github.com/MVIG-SJTU/AlphaPose)\n2. [Hao-Shu Fang and colleagues (2022). AlphaPose: Whole-Body Regional Multi-Person Pose Estimation and Tracking in Real-Time. IEEE Transactions on Pattern Analysis and Machine Intelligence.](https://doi.org/10.1109/tpami.2022.3222784)\n3. [RMPE: Regional Multi-Person Pose Estimation](https://ar5iv.labs.arxiv.org/html/1612.00137)\n4. [AlphaPose MODEL_ZOO.md](https://github.com/MVIG-SJTU/AlphaPose/blob/c60106d1/docs/MODEL_ZOO.md)\n5. [Benchmarking of 2D and 3D human pose estimation frameworks on exercise videos (IEEE)](https://ieeexplore.ieee.org/document/11429610)\n6. [A systematic survey on human pose estimation: upstream and downstream tasks, approaches, lightweight models, and prospects (Artificial Intelligence Review)](https://link.springer.com/article/10.1007/s10462-024-11060-2)\n\n---\n*Topic: Encyclopedia › Technology and the built world › Computing and digital systems › Artificial intelligence and data › Language and vision AI › Computer vision › Vision methods and geometry › Pose estimation and tracking of pose*\n\n*Initially written Sep 29, 2026 · Reviewed: Sep 30, 2026 · Edited: Sep 30, 2026; Oct 25, 2026 · Last review: Sep 30, 2026*\n\n*Copyright 2026 EdgeChat AI, a subsidiary of Biostate AI.*\n\nLicense: Edgepedia Community License 1.0, https://www.edgepedia.org/edgepedia/license\n",
 "same_as": [],
 "url": "https://www.edgepedia.org/alphapose",
 "markdown_url": "https://www.edgepedia.org/alphapose.md",
 "license": {
  "name": "Edgepedia Community License 1.0",
  "url": "https://www.edgepedia.org/edgepedia/license",
  "summary": "Free with credit, commercial use included. AI training is open to everyone. For other uses, organizations over USD 100M in revenue or 100M monthly users license separately.",
  "spdx": "LicenseRef-Edgepedia-Community-1.0"
 },
 "credit": "\"AlphaPose\", Edgepedia (EdgeChat), https://www.edgepedia.org/alphapose. Edgepedia Community License 1.0.",
 "credit_md": "\"[AlphaPose](https://www.edgepedia.org/alphapose)\", Edgepedia (EdgeChat), [https://www.edgepedia.org/alphapose](https://www.edgepedia.org/alphapose). [Edgepedia Community License 1.0](https://www.edgepedia.org/edgepedia/license).",
 "credit_html": "\"<a href=\"https://www.edgepedia.org/alphapose\">AlphaPose</a>\", Edgepedia (EdgeChat), <a href=\"https://www.edgepedia.org/alphapose\">https://www.edgepedia.org/alphapose</a>. <a href=\"https://www.edgepedia.org/edgepedia/license\">Edgepedia Community License 1.0</a>.",
 "speakable": "AlphaPose is an open-source deep learning system for multi-person human pose estimation that detects people in images and video and returns their body keypoints as skeletons."
}