---
library_name: pytorch
license: other
tags:
- bu_auto
- android
pipeline_tag: keypoint-detection

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/hrnet_pose/web-assets/model_demo.png)

# HRNetPose: Optimized for Qualcomm Devices

HRNet performs pose estimation in high-resolution representations.

This is based on the implementation of HRNetPose found [here](https://github.com/leoxiaobin/deep-high-resolution-net.pytorch).
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/hrnet_pose) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).

Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.

## Getting Started
There are two ways to deploy this model on your device:

### Option 1: Download Pre-Exported Models

Below are pre-exported model assets ready for deployment.

| Runtime | Precision | Chipset | SDK Versions | Download |
|---|---|---|---|---|
| ONNX | float | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/hrnet_pose/releases/v0.51.0/hrnet_pose-onnx-float.zip)
| ONNX | w8a16 | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/hrnet_pose/releases/v0.51.0/hrnet_pose-onnx-w8a16.zip)
| ONNX | w8a8 | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/hrnet_pose/releases/v0.51.0/hrnet_pose-onnx-w8a8.zip)
| QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/hrnet_pose/releases/v0.51.0/hrnet_pose-qnn_dlc-float.zip)
| QNN_DLC | w8a16 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/hrnet_pose/releases/v0.51.0/hrnet_pose-qnn_dlc-w8a16.zip)
| QNN_DLC | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/hrnet_pose/releases/v0.51.0/hrnet_pose-qnn_dlc-w8a8.zip)
| TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/hrnet_pose/releases/v0.51.0/hrnet_pose-tflite-float.zip)
| TFLITE | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/hrnet_pose/releases/v0.51.0/hrnet_pose-tflite-w8a8.zip)

For more device-specific assets and performance metrics, visit **[HRNetPose on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/hrnet_pose)**.


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/hrnet_pose) Python library to compile and export the model with your own:
- Custom weights (e.g., fine-tuned checkpoints)
- Custom input shapes
- Target device and runtime configurations

This option is ideal if you need to customize the model beyond the default configuration provided here.

See our repository for [HRNetPose on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/hrnet_pose) for usage instructions.

## Model Details

**Model Type:** Model_use_case.pose_estimation

**Model Stats:**
- Model checkpoint: hrnet_posenet_FP32_state_dict
- Input resolution: 256x192
- Number of parameters: 28.5M
- Model size (float): 109 MB
- Model size (w8a8): 28.1 MB

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| HRNetPose | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.217 ms | 1 - 91 MB | NPU
| HRNetPose | ONNX | float | Snapdragon® X2 Elite | 1.328 ms | 55 - 55 MB | NPU
| HRNetPose | ONNX | float | Snapdragon® X Elite | 2.625 ms | 55 - 55 MB | NPU
| HRNetPose | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 1.848 ms | 0 - 148 MB | NPU
| HRNetPose | ONNX | float | Qualcomm® QCS8550 (Proxy) | 2.466 ms | 0 - 58 MB | NPU
| HRNetPose | ONNX | float | Qualcomm® QCS9075 | 3.912 ms | 0 - 4 MB | NPU
| HRNetPose | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 1.503 ms | 0 - 93 MB | NPU
| HRNetPose | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 0.788 ms | 0 - 129 MB | NPU
| HRNetPose | ONNX | w8a16 | Snapdragon® X2 Elite | 1.013 ms | 28 - 28 MB | NPU
| HRNetPose | ONNX | w8a16 | Snapdragon® X Elite | 1.987 ms | 28 - 28 MB | NPU
| HRNetPose | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 1.255 ms | 0 - 199 MB | NPU
| HRNetPose | ONNX | w8a16 | Qualcomm® QCS6490 | 480.354 ms | 30 - 34 MB | CPU
| HRNetPose | ONNX | w8a16 | Qualcomm® QCS8550 (Proxy) | 1.761 ms | 0 - 38 MB | NPU
| HRNetPose | ONNX | w8a16 | Qualcomm® QCS9075 | 1.989 ms | 0 - 3 MB | NPU
| HRNetPose | ONNX | w8a16 | Qualcomm® QCM6690 | 223.1 ms | 30 - 48 MB | CPU
| HRNetPose | ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 0.972 ms | 0 - 124 MB | NPU
| HRNetPose | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 214.16 ms | 26 - 41 MB | CPU
| HRNetPose | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.806 ms | 0 - 122 MB | NPU
| HRNetPose | ONNX | w8a8 | Snapdragon® X2 Elite | 0.77 ms | 30 - 30 MB | NPU
| HRNetPose | ONNX | w8a8 | Snapdragon® X Elite | 1.632 ms | 28 - 28 MB | NPU
| HRNetPose | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.046 ms | 0 - 186 MB | NPU
| HRNetPose | ONNX | w8a8 | Qualcomm® QCS6490 | 89.757 ms | 8 - 68 MB | CPU
| HRNetPose | ONNX | w8a8 | Qualcomm® QCS8550 (Proxy) | 1.513 ms | 0 - 33 MB | NPU
| HRNetPose | ONNX | w8a8 | Qualcomm® QCS9075 | 1.542 ms | 0 - 3 MB | NPU
| HRNetPose | ONNX | w8a8 | Qualcomm® QCM6690 | 60.819 ms | 9 - 30 MB | CPU
| HRNetPose | ONNX | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.912 ms | 0 - 114 MB | NPU
| HRNetPose | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 58.162 ms | 10 - 30 MB | CPU
| HRNetPose | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.266 ms | 1 - 77 MB | NPU
| HRNetPose | QNN_DLC | float | Snapdragon® X2 Elite | 1.689 ms | 1 - 1 MB | NPU
| HRNetPose | QNN_DLC | float | Snapdragon® X Elite | 2.937 ms | 1 - 1 MB | NPU
| HRNetPose | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 1.991 ms | 0 - 117 MB | NPU
| HRNetPose | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 14.17 ms | 1 - 73 MB | NPU
| HRNetPose | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 2.776 ms | 1 - 2 MB | NPU
| HRNetPose | QNN_DLC | float | Qualcomm® SA8775P | 4.291 ms | 1 - 75 MB | NPU
| HRNetPose | QNN_DLC | float | Qualcomm® QCS9075 | 4.105 ms | 1 - 3 MB | NPU
| HRNetPose | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 5.016 ms | 0 - 103 MB | NPU
| HRNetPose | QNN_DLC | float | Qualcomm® SA7255P | 14.17 ms | 1 - 73 MB | NPU
| HRNetPose | QNN_DLC | float | Qualcomm® SA8295P | 4.581 ms | 1 - 61 MB | NPU
| HRNetPose | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 1.561 ms | 0 - 73 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 0.809 ms | 0 - 101 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 1.307 ms | 0 - 0 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Snapdragon® X Elite | 2.141 ms | 0 - 0 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 1.368 ms | 0 - 147 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Qualcomm® QCS6490 | 6.463 ms | 0 - 2 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Qualcomm® QCS8275 (Proxy) | 5.154 ms | 0 - 100 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Qualcomm® QCS8550 (Proxy) | 1.867 ms | 0 - 2 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Qualcomm® SA8775P | 2.255 ms | 0 - 102 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Qualcomm® QCS9075 | 2.179 ms | 2 - 4 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Qualcomm® QCM6690 | 19.73 ms | 0 - 218 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Qualcomm® QCS8450 (Proxy) | 2.727 ms | 0 - 149 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Qualcomm® SA7255P | 5.154 ms | 0 - 100 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Qualcomm® SA8295P | 3.071 ms | 0 - 99 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 1.019 ms | 0 - 100 MB | NPU
| HRNetPose | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 2.548 ms | 0 - 101 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.546 ms | 0 - 94 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 0.719 ms | 0 - 0 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Snapdragon® X Elite | 1.267 ms | 0 - 0 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.824 ms | 0 - 133 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Qualcomm® QCS6490 | 3.73 ms | 0 - 2 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Qualcomm® QCS8275 (Proxy) | 2.837 ms | 0 - 90 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Qualcomm® QCS8550 (Proxy) | 1.127 ms | 0 - 2 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Qualcomm® SA8775P | 1.502 ms | 0 - 91 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Qualcomm® QCS9075 | 1.282 ms | 0 - 2 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Qualcomm® QCM6690 | 10.449 ms | 0 - 213 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Qualcomm® QCS8450 (Proxy) | 1.734 ms | 0 - 135 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Qualcomm® SA7255P | 2.837 ms | 0 - 90 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Qualcomm® SA8295P | 1.952 ms | 0 - 88 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.642 ms | 0 - 92 MB | NPU
| HRNetPose | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.513 ms | 0 - 90 MB | NPU
| HRNetPose | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.277 ms | 0 - 113 MB | NPU
| HRNetPose | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 1.988 ms | 0 - 184 MB | NPU
| HRNetPose | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 14.166 ms | 0 - 111 MB | NPU
| HRNetPose | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 2.766 ms | 0 - 2 MB | NPU
| HRNetPose | TFLITE | float | Qualcomm® SA8775P | 4.338 ms | 0 - 111 MB | NPU
| HRNetPose | TFLITE | float | Qualcomm® QCS9075 | 4.147 ms | 0 - 58 MB | NPU
| HRNetPose | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 5.029 ms | 0 - 174 MB | NPU
| HRNetPose | TFLITE | float | Qualcomm® SA7255P | 14.166 ms | 0 - 111 MB | NPU
| HRNetPose | TFLITE | float | Qualcomm® SA8295P | 4.624 ms | 0 - 103 MB | NPU
| HRNetPose | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 1.548 ms | 0 - 112 MB | NPU
| HRNetPose | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.513 ms | 0 - 91 MB | NPU
| HRNetPose | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.712 ms | 0 - 140 MB | NPU
| HRNetPose | TFLITE | w8a8 | Qualcomm® QCS6490 | 3.444 ms | 0 - 30 MB | NPU
| HRNetPose | TFLITE | w8a8 | Qualcomm® QCS8275 (Proxy) | 2.547 ms | 0 - 87 MB | NPU
| HRNetPose | TFLITE | w8a8 | Qualcomm® QCS8550 (Proxy) | 0.952 ms | 0 - 3 MB | NPU
| HRNetPose | TFLITE | w8a8 | Qualcomm® SA8775P | 1.355 ms | 0 - 89 MB | NPU
| HRNetPose | TFLITE | w8a8 | Qualcomm® QCS9075 | 1.081 ms | 0 - 30 MB | NPU
| HRNetPose | TFLITE | w8a8 | Qualcomm® QCM6690 | 9.629 ms | 0 - 207 MB | NPU
| HRNetPose | TFLITE | w8a8 | Qualcomm® QCS8450 (Proxy) | 1.523 ms | 0 - 142 MB | NPU
| HRNetPose | TFLITE | w8a8 | Qualcomm® SA7255P | 2.547 ms | 0 - 87 MB | NPU
| HRNetPose | TFLITE | w8a8 | Qualcomm® SA8295P | 1.792 ms | 0 - 85 MB | NPU
| HRNetPose | TFLITE | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.586 ms | 0 - 88 MB | NPU
| HRNetPose | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.355 ms | 0 - 85 MB | NPU

## License
* The license for the original implementation of HRNetPose can be found
  [here](https://github.com/leoxiaobin/deep-high-resolution-net.pytorch/blob/master/LICENSE).

## References
* [Deep High-Resolution Representation Learning for Human Pose Estimation](https://arxiv.org/abs/1902.09212)
* [Source Model Implementation](https://github.com/leoxiaobin/deep-high-resolution-net.pytorch)

## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).