LiteHRNet: Optimized for Qualcomm Devices

LiteHRNet is a machine learning model that detects human pose and returns a location and confidence for each of 17 joints.

This is based on the implementation of LiteHRNet found here. This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.

Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up 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 2.45, ONNX Runtime 1.27.1	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit LiteHRNet on Qualcomm® AI Hub.

Option 2: Export with Custom Configurations

Use the Qualcomm® AI Hub Models 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 LiteHRNet on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.pose_estimation

Model Stats:

• Input resolution: 256x192
• Model size (float): 4.49 MB
• Number of parameters: 1.11M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
LiteHRNet	ONNX	float	Snapdragon® X2 Elite	2.878 ms	2 - 2 MB	NPU
LiteHRNet	ONNX	float	Snapdragon® X Elite	5.735 ms	5 - 5 MB	NPU
LiteHRNet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	3.108 ms	0 - 123 MB	NPU
LiteHRNet	ONNX	float	Snapdragon® 8 Gen 1 Mobile	6.316 ms	1 - 122 MB	NPU
LiteHRNet	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	4.465 ms	1 - 5 MB	NPU
LiteHRNet	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	5.419 ms	1 - 114 MB	NPU
LiteHRNet	ONNX	float	Qualcomm® QCS8450	6.316 ms	1 - 122 MB	NPU
LiteHRNet	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	5.773 ms	1 - 4 MB	NPU
LiteHRNet	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	5.735 ms	5 - 5 MB	NPU
LiteHRNet	ONNX	float	Qualcomm® Dragonwing™ Q-8750	2.881 ms	0 - 98 MB	NPU
LiteHRNet	ONNX	float	Snapdragon® 8 Elite Mobile	2.881 ms	0 - 98 MB	NPU
LiteHRNet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	2.772 ms	0 - 100 MB	NPU
LiteHRNet	QNN_DLC	float	Snapdragon® X2 Elite	1.292 ms	1 - 1 MB	NPU
LiteHRNet	QNN_DLC	float	Snapdragon® X Elite	2.442 ms	1 - 1 MB	NPU
LiteHRNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	1.393 ms	0 - 104 MB	NPU
LiteHRNet	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	2.951 ms	0 - 101 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	2.192 ms	1 - 4 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	5.034 ms	1 - 78 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	2.117 ms	1 - 91 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® SA8775P	2.719 ms	1 - 80 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® SA8650P	2.719 ms	1 - 80 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® SA8255P	2.719 ms	1 - 80 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® QCS8450	2.951 ms	0 - 101 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	2.908 ms	1 - 3 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	2.442 ms	1 - 1 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	1.075 ms	0 - 80 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® SA7255P	5.034 ms	1 - 78 MB	NPU
LiteHRNet	QNN_DLC	float	Qualcomm® SA8295P	3.482 ms	0 - 77 MB	NPU
LiteHRNet	QNN_DLC	float	Snapdragon® 8 Elite Mobile	1.075 ms	0 - 80 MB	NPU
LiteHRNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.913 ms	1 - 83 MB	NPU
LiteHRNet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	2.686 ms	0 - 149 MB	NPU
LiteHRNet	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	5.368 ms	1 - 138 MB	NPU
LiteHRNet	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	4.23 ms	1 - 12 MB	NPU
LiteHRNet	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	8.834 ms	1 - 116 MB	NPU
LiteHRNet	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	4.262 ms	0 - 3 MB	NPU
LiteHRNet	TFLITE	float	Qualcomm® SA8775P	5.35 ms	1 - 116 MB	NPU
LiteHRNet	TFLITE	float	Qualcomm® SA8650P	5.35 ms	1 - 116 MB	NPU
LiteHRNet	TFLITE	float	Qualcomm® SA8255P	5.35 ms	1 - 116 MB	NPU
LiteHRNet	TFLITE	float	Qualcomm® QCS8450	5.368 ms	1 - 138 MB	NPU
LiteHRNet	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	4.924 ms	1 - 11 MB	NPU
LiteHRNet	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	2.251 ms	0 - 119 MB	NPU
LiteHRNet	TFLITE	float	Qualcomm® SA7255P	8.834 ms	1 - 116 MB	NPU
LiteHRNet	TFLITE	float	Qualcomm® SA8295P	6.4 ms	1 - 113 MB	NPU
LiteHRNet	TFLITE	float	Snapdragon® 8 Elite Mobile	2.251 ms	0 - 119 MB	NPU
LiteHRNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	2.051 ms	0 - 110 MB	NPU

License

• The license for the original implementation of LiteHRNet can be found here.

References

• Lite-HRNet: A Lightweight High-Resolution Network
• Source Model Implementation

Community

• Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
• For questions or feedback please reach out to us.