SixDRepNet: Optimized for Qualcomm Devices

6DRepNet predicts head pose (pitch, yaw, roll) from a face image using a RepVGG-B1g2 backbone and a continuous 6D rotation representation, achieving robust and accurate head pose estimation.

This is based on the implementation of SixDRepNet 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 2.42, ONNX Runtime 1.24.3	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 SixDRepNet 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 SixDRepNet on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.pose_estimation

Model Stats:

• Input resolution: 224x224
• Number of parameters: 15.3M
• Model size (float): 58.4 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
face_detector	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	1.483 ms	0 - 166 MB	NPU
face_detector	ONNX	float	Snapdragon® 8 Elite Mobile	1.863 ms	3 - 166 MB	NPU
face_detector	ONNX	float	Snapdragon® X2 Elite	1.581 ms	7 - 7 MB	NPU
face_detector	ONNX	float	Snapdragon® X Elite	3.803 ms	7 - 7 MB	NPU
face_detector	ONNX	float	Snapdragon® X Elite	3.803 ms	7 - 7 MB	NPU
face_detector	ONNX	float	Snapdragon® 8 Gen 3 Mobile	2.201 ms	3 - 169 MB	NPU
face_detector	ONNX	float	Qualcomm® QCS8550 (Proxy)	3.469 ms	0 - 4 MB	NPU
face_detector	ONNX	float	Qualcomm® QCS9075	5.366 ms	4 - 12 MB	NPU
face_detector	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	1.863 ms	3 - 166 MB	NPU
face_detector	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	5.42 ms	5 - 159 MB	NPU
face_detector	QNN_DLC	float	Snapdragon® 8 Elite Mobile	6.896 ms	0 - 150 MB	NPU
face_detector	QNN_DLC	float	Snapdragon® X2 Elite	5.876 ms	5 - 5 MB	NPU
face_detector	QNN_DLC	float	Snapdragon® X Elite	16.475 ms	5 - 5 MB	NPU
face_detector	QNN_DLC	float	Snapdragon® X Elite	16.475 ms	5 - 5 MB	NPU
face_detector	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	9.278 ms	5 - 173 MB	NPU
face_detector	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	28.312 ms	1 - 151 MB	NPU
face_detector	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	15.67 ms	5 - 8 MB	NPU
face_detector	QNN_DLC	float	Qualcomm® SA8775P	16.54 ms	1 - 153 MB	NPU
face_detector	QNN_DLC	float	Qualcomm® SA8775P	16.54 ms	1 - 153 MB	NPU
face_detector	QNN_DLC	float	Qualcomm® SA8775P	16.54 ms	1 - 153 MB	NPU
face_detector	QNN_DLC	float	Qualcomm® QCS9075	19.498 ms	5 - 12 MB	NPU
face_detector	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	24.733 ms	5 - 179 MB	NPU
face_detector	QNN_DLC	float	Qualcomm® SA7255P	28.312 ms	1 - 151 MB	NPU
face_detector	QNN_DLC	float	Qualcomm® SA8295P	20.592 ms	0 - 150 MB	NPU
face_detector	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	6.896 ms	0 - 150 MB	NPU
face_detector	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	5.482 ms	1 - 155 MB	NPU
face_detector	TFLITE	float	Snapdragon® 8 Elite Mobile	6.923 ms	0 - 150 MB	NPU
face_detector	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	9.272 ms	1 - 171 MB	NPU
face_detector	TFLITE	float	Qualcomm® QCS8275 (Proxy)	28.26 ms	1 - 150 MB	NPU
face_detector	TFLITE	float	Qualcomm® QCS8550 (Proxy)	15.666 ms	1 - 3 MB	NPU
face_detector	TFLITE	float	Qualcomm® SA8775P	16.531 ms	1 - 152 MB	NPU
face_detector	TFLITE	float	Qualcomm® SA8775P	16.531 ms	1 - 152 MB	NPU
face_detector	TFLITE	float	Qualcomm® SA8775P	16.531 ms	1 - 152 MB	NPU
face_detector	TFLITE	float	Qualcomm® QCS9075	19.66 ms	1 - 10 MB	NPU
face_detector	TFLITE	float	Qualcomm® QCS8450 (Proxy)	24.453 ms	1 - 171 MB	NPU
face_detector	TFLITE	float	Qualcomm® SA7255P	28.26 ms	1 - 150 MB	NPU
face_detector	TFLITE	float	Qualcomm® SA8295P	20.653 ms	1 - 152 MB	NPU
face_detector	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	6.923 ms	0 - 150 MB	NPU
pose_estimator	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	1.334 ms	0 - 24 MB	NPU
pose_estimator	ONNX	float	Snapdragon® 8 Elite Mobile	1.619 ms	0 - 22 MB	NPU
pose_estimator	ONNX	float	Snapdragon® X2 Elite	1.318 ms	75 - 75 MB	NPU
pose_estimator	ONNX	float	Snapdragon® X Elite	2.633 ms	75 - 75 MB	NPU
pose_estimator	ONNX	float	Snapdragon® X Elite	2.633 ms	75 - 75 MB	NPU
pose_estimator	ONNX	float	Snapdragon® 8 Gen 3 Mobile	2.059 ms	0 - 35 MB	NPU
pose_estimator	ONNX	float	Qualcomm® QCS8550 (Proxy)	2.627 ms	1 - 2 MB	NPU
pose_estimator	ONNX	float	Qualcomm® QCS9075	4.551 ms	0 - 4 MB	NPU
pose_estimator	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	1.619 ms	0 - 22 MB	NPU
pose_estimator	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	1.417 ms	1 - 29 MB	NPU
pose_estimator	QNN_DLC	float	Snapdragon® 8 Elite Mobile	1.682 ms	0 - 26 MB	NPU
pose_estimator	QNN_DLC	float	Snapdragon® X2 Elite	1.487 ms	1 - 1 MB	NPU
pose_estimator	QNN_DLC	float	Snapdragon® X Elite	2.864 ms	1 - 1 MB	NPU
pose_estimator	QNN_DLC	float	Snapdragon® X Elite	2.864 ms	1 - 1 MB	NPU
pose_estimator	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	2.235 ms	1 - 40 MB	NPU
pose_estimator	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	17.816 ms	1 - 24 MB	NPU
pose_estimator	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	2.782 ms	1 - 2 MB	NPU
pose_estimator	QNN_DLC	float	Qualcomm® SA8775P	4.813 ms	1 - 26 MB	NPU
pose_estimator	QNN_DLC	float	Qualcomm® SA8775P	4.813 ms	1 - 26 MB	NPU
pose_estimator	QNN_DLC	float	Qualcomm® SA8775P	4.813 ms	1 - 26 MB	NPU
pose_estimator	QNN_DLC	float	Qualcomm® QCS9075	4.904 ms	3 - 5 MB	NPU
pose_estimator	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	6.543 ms	0 - 39 MB	NPU
pose_estimator	QNN_DLC	float	Qualcomm® SA7255P	17.816 ms	1 - 24 MB	NPU
pose_estimator	QNN_DLC	float	Qualcomm® SA8295P	5.398 ms	1 - 23 MB	NPU
pose_estimator	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	1.682 ms	0 - 26 MB	NPU
pose_estimator	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	1.374 ms	0 - 28 MB	NPU
pose_estimator	TFLITE	float	Snapdragon® 8 Elite Mobile	1.701 ms	0 - 30 MB	NPU
pose_estimator	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	2.265 ms	0 - 46 MB	NPU
pose_estimator	TFLITE	float	Qualcomm® QCS8275 (Proxy)	17.398 ms	0 - 26 MB	NPU
pose_estimator	TFLITE	float	Qualcomm® QCS8550 (Proxy)	2.849 ms	0 - 3 MB	NPU
pose_estimator	TFLITE	float	Qualcomm® SA8775P	4.813 ms	0 - 27 MB	NPU
pose_estimator	TFLITE	float	Qualcomm® SA8775P	4.813 ms	0 - 27 MB	NPU
pose_estimator	TFLITE	float	Qualcomm® SA8775P	4.813 ms	0 - 27 MB	NPU
pose_estimator	TFLITE	float	Qualcomm® QCS9075	4.74 ms	0 - 78 MB	NPU
pose_estimator	TFLITE	float	Qualcomm® QCS8450 (Proxy)	6.502 ms	0 - 43 MB	NPU
pose_estimator	TFLITE	float	Qualcomm® SA7255P	17.398 ms	0 - 26 MB	NPU
pose_estimator	TFLITE	float	Qualcomm® SA8295P	5.349 ms	0 - 28 MB	NPU
pose_estimator	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	1.701 ms	0 - 30 MB	NPU

License

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

References

• 6D Rotation Representation for Unconstrained Head Pose Estimation
• 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.