HRNetFace: Optimized for Qualcomm Devices

Detects attributes (liveness, eye closeness, mask presence, glasses presence, sunglasses presence) that apply to a given face.

This is based on the implementation of HRNetFace 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 1.25.0	Download
ONNX	w8a8	Universal	QAIRT 2.45, ONNX Runtime 1.25.0	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	w8a8	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download
TFLITE	w8a8	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit HRNetFace 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 HRNetFace on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.object_detection

Model Stats:

• Model checkpoint: HR18-COFW.pth
• Input resolution: 256x256
• Number of parameters: 9.68M
• Model size (float): 36.87MB
• Model size (w8a8): 17.7 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
HRNetFace	ONNX	float	Snapdragon® X2 Elite	1.543 ms	212 - 212 MB	NPU
HRNetFace	ONNX	float	Snapdragon® X Elite	3.146 ms	149 - 149 MB	NPU
HRNetFace	ONNX	float	Snapdragon® 8 Gen 3 Mobile	2.157 ms	1 - 136 MB	NPU
HRNetFace	ONNX	float	Snapdragon® 8 Gen 1 Mobile	4.53 ms	1 - 121 MB	NPU
HRNetFace	ONNX	float	Qualcomm® QCS8550 (Proxy)	3.036 ms	0 - 4 MB	NPU
HRNetFace	ONNX	float	Qualcomm® QCS8450	4.53 ms	1 - 121 MB	NPU
HRNetFace	ONNX	float	Snapdragon® 8 Elite Mobile	1.652 ms	0 - 92 MB	NPU
HRNetFace	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	1.331 ms	1 - 88 MB	NPU
HRNetFace	ONNX	float	Qualcomm® QCS9075	4.766 ms	2 - 47 MB	NPU
HRNetFace	ONNX	float	Qualcomm® QCS8750	1.652 ms	0 - 92 MB	NPU
HRNetFace	ONNX	float	Qualcomm® QCS7181	3.146 ms	149 - 149 MB	NPU
HRNetFace	ONNX	w8a8	Snapdragon® X2 Elite	0.665 ms	213 - 213 MB	NPU
HRNetFace	ONNX	w8a8	Snapdragon® X Elite	1.329 ms	149 - 149 MB	NPU
HRNetFace	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	0.88 ms	0 - 147 MB	NPU
HRNetFace	ONNX	w8a8	Snapdragon® 8 Gen 1 Mobile	1.734 ms	0 - 147 MB	NPU
HRNetFace	ONNX	w8a8	Qualcomm® QCS6490	3.526 ms	0 - 45 MB	NPU
HRNetFace	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	1.299 ms	0 - 20 MB	NPU
HRNetFace	ONNX	w8a8	Qualcomm® QCS8450	1.734 ms	0 - 147 MB	NPU
HRNetFace	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.591 ms	0 - 106 MB	NPU
HRNetFace	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	1.598 ms	0 - 111 MB	NPU
HRNetFace	ONNX	w8a8	Qualcomm® QCM6690	10.036 ms	0 - 230 MB	NPU
HRNetFace	ONNX	w8a8	Qualcomm® QCS9075	1.521 ms	0 - 45 MB	NPU
HRNetFace	ONNX	w8a8	Snapdragon® 8 Elite Mobile	0.7 ms	0 - 109 MB	NPU
HRNetFace	ONNX	w8a8	Qualcomm® QCS7790	1.598 ms	0 - 111 MB	NPU
HRNetFace	ONNX	w8a8	Qualcomm® QCS8750	0.7 ms	0 - 109 MB	NPU
HRNetFace	ONNX	w8a8	Qualcomm® QCS7181	1.329 ms	149 - 149 MB	NPU
HRNetFace	QNN_DLC	float	Snapdragon® X2 Elite	1.937 ms	1 - 1 MB	NPU
HRNetFace	QNN_DLC	float	Snapdragon® X Elite	3.733 ms	1 - 1 MB	NPU
HRNetFace	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	2.327 ms	0 - 120 MB	NPU
HRNetFace	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	5.509 ms	0 - 103 MB	NPU
HRNetFace	QNN_DLC	float	Qualcomm® QCS8275	15.456 ms	1 - 75 MB	NPU
HRNetFace	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	3.279 ms	1 - 3 MB	NPU
HRNetFace	QNN_DLC	float	Qualcomm® QCS8450	5.509 ms	0 - 103 MB	NPU
HRNetFace	QNN_DLC	float	Snapdragon® 8 Elite Mobile	1.752 ms	1 - 80 MB	NPU
HRNetFace	QNN_DLC	float	Qualcomm® SA8295P	5.66 ms	1 - 62 MB	NPU
HRNetFace	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	1.399 ms	1 - 81 MB	NPU
HRNetFace	QNN_DLC	float	Qualcomm® SA7255P	15.456 ms	1 - 75 MB	NPU
HRNetFace	QNN_DLC	float	Qualcomm® QCS9075	5.026 ms	1 - 4 MB	NPU
HRNetFace	QNN_DLC	float	Qualcomm® QCS8750	1.752 ms	1 - 80 MB	NPU
HRNetFace	QNN_DLC	float	Qualcomm® QCS7181	3.733 ms	1 - 1 MB	NPU
HRNetFace	QNN_DLC	w8a8	Snapdragon® X2 Elite	0.848 ms	0 - 0 MB	NPU
HRNetFace	QNN_DLC	w8a8	Snapdragon® X Elite	1.539 ms	0 - 0 MB	NPU
HRNetFace	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	0.902 ms	0 - 108 MB	NPU
HRNetFace	QNN_DLC	w8a8	Snapdragon® 8 Gen 1 Mobile	1.849 ms	0 - 108 MB	NPU
HRNetFace	QNN_DLC	w8a8	Qualcomm® QCS6490	3.829 ms	0 - 2 MB	NPU
HRNetFace	QNN_DLC	w8a8	Qualcomm® QCS8275	3.272 ms	0 - 71 MB	NPU
HRNetFace	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	1.331 ms	0 - 2 MB	NPU
HRNetFace	QNN_DLC	w8a8	Qualcomm® QCS8450	1.849 ms	0 - 108 MB	NPU
HRNetFace	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.542 ms	0 - 74 MB	NPU
HRNetFace	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	1.613 ms	0 - 79 MB	NPU
HRNetFace	QNN_DLC	w8a8	Qualcomm® QCM6690	10.617 ms	0 - 200 MB	NPU
HRNetFace	QNN_DLC	w8a8	Qualcomm® QCS9075	1.611 ms	2 - 4 MB	NPU
HRNetFace	QNN_DLC	w8a8	Qualcomm® SA7255P	3.272 ms	0 - 71 MB	NPU
HRNetFace	QNN_DLC	w8a8	Snapdragon® 8 Elite Mobile	0.676 ms	0 - 69 MB	NPU
HRNetFace	QNN_DLC	w8a8	Qualcomm® SA8295P	2.215 ms	0 - 70 MB	NPU
HRNetFace	QNN_DLC	w8a8	Qualcomm® QCS7790	1.613 ms	0 - 79 MB	NPU
HRNetFace	QNN_DLC	w8a8	Qualcomm® QCS8750	0.676 ms	0 - 69 MB	NPU
HRNetFace	QNN_DLC	w8a8	Qualcomm® QCS7181	1.539 ms	0 - 0 MB	NPU
HRNetFace	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	2.339 ms	0 - 139 MB	NPU
HRNetFace	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	5.486 ms	0 - 125 MB	NPU
HRNetFace	TFLITE	float	Qualcomm® QCS8275	15.477 ms	1 - 91 MB	NPU
HRNetFace	TFLITE	float	Qualcomm® QCS8550 (Proxy)	3.351 ms	0 - 3 MB	NPU
HRNetFace	TFLITE	float	Qualcomm® SA8775P	38.193 ms	1 - 43 MB	GPU
HRNetFace	TFLITE	float	Qualcomm® SA8650P	38.193 ms	1 - 43 MB	GPU
HRNetFace	TFLITE	float	Qualcomm® SA8255P	38.193 ms	1 - 43 MB	GPU
HRNetFace	TFLITE	float	Qualcomm® QCS8450	5.486 ms	0 - 125 MB	NPU
HRNetFace	TFLITE	float	Snapdragon® 8 Elite Mobile	1.758 ms	0 - 94 MB	NPU
HRNetFace	TFLITE	float	Qualcomm® SA8295P	5.652 ms	1 - 74 MB	NPU
HRNetFace	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	1.391 ms	0 - 89 MB	NPU
HRNetFace	TFLITE	float	Qualcomm® SA7255P	15.477 ms	1 - 91 MB	NPU
HRNetFace	TFLITE	float	Qualcomm® QCS9075	5.044 ms	0 - 35 MB	NPU
HRNetFace	TFLITE	float	Qualcomm® QCS8750	1.758 ms	0 - 94 MB	NPU
HRNetFace	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	0.737 ms	0 - 112 MB	NPU
HRNetFace	TFLITE	w8a8	Snapdragon® 8 Gen 1 Mobile	1.489 ms	0 - 113 MB	NPU
HRNetFace	TFLITE	w8a8	Qualcomm® QCS6490	3.221 ms	0 - 18 MB	NPU
HRNetFace	TFLITE	w8a8	Qualcomm® QCS8275	2.893 ms	0 - 71 MB	NPU
HRNetFace	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	1.092 ms	0 - 3 MB	NPU
HRNetFace	TFLITE	w8a8	Qualcomm® SA8775P	39.03 ms	2 - 48 MB	GPU
HRNetFace	TFLITE	w8a8	Qualcomm® SA8650P	39.03 ms	2 - 48 MB	GPU
HRNetFace	TFLITE	w8a8	Qualcomm® SA8255P	39.03 ms	2 - 48 MB	GPU
HRNetFace	TFLITE	w8a8	Qualcomm® QCS8450	1.489 ms	0 - 113 MB	NPU
HRNetFace	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.461 ms	0 - 72 MB	NPU
HRNetFace	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	1.353 ms	0 - 75 MB	NPU
HRNetFace	TFLITE	w8a8	Qualcomm® QCM6690	9.849 ms	0 - 193 MB	NPU
HRNetFace	TFLITE	w8a8	Qualcomm® QCS9075	1.323 ms	0 - 18 MB	NPU
HRNetFace	TFLITE	w8a8	Qualcomm® SA7255P	2.893 ms	0 - 71 MB	NPU
HRNetFace	TFLITE	w8a8	Snapdragon® 8 Elite Mobile	0.577 ms	0 - 68 MB	NPU
HRNetFace	TFLITE	w8a8	Qualcomm® SA8295P	2.005 ms	0 - 69 MB	NPU
HRNetFace	TFLITE	w8a8	Qualcomm® QCS7790	1.353 ms	0 - 75 MB	NPU
HRNetFace	TFLITE	w8a8	Qualcomm® QCS8750	0.577 ms	0 - 68 MB	NPU

License

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

References

• Deep High-Resolution Representation Learning for Visual Recognition
• 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.