Project AURIONAI Use-Cases Research & Innovation
visionfamily

HRNet

High-resolution networks that keep high-resolution features through the whole network, used for human pose estimation, segmentation and as backbones.

University of Science and Technology of China (USTC)University of Science and Technology of China (USTC)ChinaMicrosoft ResearchMicrosoft ResearchUSA
University of Science and Technology of China (USTC)
Levelfamily
Classvision
Parametersnot in the notes
Contextnot in the notes
Released25 Feb 2019
Opennessopen source
Runs onmobile or NPU, CPU server
Statusactive

Lineage

2019202020212022HRNet (pose)HRNet (pose), Feb 2019HRNetV2HRNetV2, Aug 2019HigherHRNetHigherHRNet, Aug 2019Lite-HRNetLite-HRNet, Apr 2021HRFormerHRFormer, Oct 2021
Peking UniversityHRFormer belongs to this familyvision18 Oct 2021
Huazhong University of Science and TechnologyLite-HRNet belongs to this familyvision13 Apr 2021
University of Illinois Urbana-ChampaignHigherHRNet belongs to this familyvision27 Aug 2019
Microsoft ResearchHRNetV2 belongs to this familyvision20 Aug 2019
University of Science and Technology of China (USTC)HRNet (pose) belongs to this familyvision25 Feb 2019
Where this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. Confidence and evidence are the notes' own.

R&D Topics

Image classification backbone architectures (CNN and vision transformer)produced byHuman pose estimationproduced by

Identity

Developer
University of Science and Technology of China (USTC), Microsoft Research
Family
not in the notes
Released
25 Feb 2019 sourcesannouncement page
Weights published
–
Record checked
12 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.

Architecture and Size

Architecture
CNN
Total parameters
– sources
Active parameters
–
Checkpoint parameters
–
Note
HRNet-W32 and W48 for pose; HRNetV2 for classification and segmentation (arXiv 1908.07919); HigherHRNet; Lite-HRNet; HRFormer.
Context length
not in the notes
Input
image
Output
not in the notes
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).

Openness and Access

Openness
open source
Licence
MIT License sources
Access routes
open weights download
Where this comes from
The licence file governs over registry tags and aggregator labels.

Deployment

Runs on
mobile or NPU, CPU server
Footprint
Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.
Runtimes
not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.

Indonesia

Indonesian support
not stated
Regional languages
not in the notes
Jakarta hosting
unknown
Languages
not in the notes
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.

Benchmark Results

The notes record no benchmark result for this model.

Industry Use Cases

No industry use case links this model.

Sources

  1. arxiv.org/abs/1902.09212 https://arxiv.org/abs/1902.09212
  2. github.com/leoxiaobin/deep-high-resolution-net.pytorch https://github.com/leoxiaobin/deep-high-resolution-net.pytorch
  3. aihub.qualcomm.com/models/hrnet_pose https://aihub.qualcomm.com/models/hrnet_pose
  4. aihub.qualcomm.com/models/litehrnet https://aihub.qualcomm.com/models/litehrnet
  5. github.com/HRNet/Lite-HRNet https://github.com/HRNet/Lite-HRNet
  6. github.com/openvinotoolkit/open_model_zoo/tree/master/mo... https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/higher-hrnet-w32-human-pose-estimation
  7. modelscope.cn/models/iic/cv_hrnetv2w32_body-2d-keypoints_im... https://www.modelscope.cn/models/iic/cv_hrnetv2w32_body-2d-keypoints_image
  8. arxiv.org/html/1902.09212 https://arxiv.org/html/1902.09212

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "HRNet", https://aurion.semai.id/models/hrnet/, records of October 2026.