MediaPipe-Hand-Detection: Optimized for Qualcomm Devices

The MediaPipe Hand Landmark Detector is a machine learning pipeline that predicts bounding boxes and pose skeletons of hands in an image.

This is based on the implementation of MediaPipe-Hand-Detection 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 2.43	Download
TFLITE	float	Universal	QAIRT 2.43, TFLite 2.19.1	Download

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

Model Details

Model Type: Model_use_case.object_detection

Model Stats:

• Input resolution: 256x256
• Number of parameters (hand_detector): 1.76M
• Model size (hand_detector) (float): 6.75 MB
• Number of parameters (hand_landmark_detector): 2.01M
• Model size (hand_landmark_detector) (float): 7.70 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
hand_detector	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.42 ms	0 - 47 MB	NPU
hand_detector	ONNX	float	Snapdragon® X2 Elite	0.492 ms	0 - 0 MB	NPU
hand_detector	ONNX	float	Snapdragon® X Elite	1.011 ms	3 - 3 MB	NPU
hand_detector	ONNX	float	Snapdragon® 8 Gen 3 Mobile	0.594 ms	0 - 71 MB	NPU
hand_detector	ONNX	float	Qualcomm® QCS8550 (Proxy)	0.837 ms	0 - 7 MB	NPU
hand_detector	ONNX	float	Qualcomm® QCS9075	1.265 ms	1 - 3 MB	NPU
hand_detector	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	0.493 ms	0 - 48 MB	NPU
hand_detector	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.377 ms	1 - 43 MB	NPU
hand_detector	QNN_DLC	float	Snapdragon® X2 Elite	0.651 ms	1 - 1 MB	NPU
hand_detector	QNN_DLC	float	Snapdragon® X Elite	0.903 ms	1 - 1 MB	NPU
hand_detector	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	0.53 ms	0 - 59 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	3.774 ms	0 - 40 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	0.736 ms	1 - 2 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® SA8775P	1.287 ms	1 - 43 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® QCS9075	1.13 ms	1 - 3 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	1.352 ms	0 - 53 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® SA7255P	3.774 ms	0 - 40 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® SA8295P	1.688 ms	1 - 32 MB	NPU
hand_detector	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	0.445 ms	0 - 43 MB	NPU
hand_detector	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.381 ms	0 - 42 MB	NPU
hand_detector	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	0.535 ms	0 - 58 MB	NPU
hand_detector	TFLITE	float	Qualcomm® QCS8275 (Proxy)	3.808 ms	0 - 38 MB	NPU
hand_detector	TFLITE	float	Qualcomm® QCS8550 (Proxy)	0.734 ms	0 - 104 MB	NPU
hand_detector	TFLITE	float	Qualcomm® SA8775P	1.327 ms	0 - 42 MB	NPU
hand_detector	TFLITE	float	Qualcomm® QCS9075	1.138 ms	0 - 7 MB	NPU
hand_detector	TFLITE	float	Qualcomm® QCS8450 (Proxy)	1.365 ms	0 - 51 MB	NPU
hand_detector	TFLITE	float	Qualcomm® SA7255P	3.808 ms	0 - 38 MB	NPU
hand_detector	TFLITE	float	Qualcomm® SA8295P	1.714 ms	0 - 30 MB	NPU
hand_detector	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	0.451 ms	0 - 37 MB	NPU
hand_landmark_detector	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.562 ms	0 - 43 MB	NPU
hand_landmark_detector	ONNX	float	Snapdragon® X2 Elite	0.723 ms	6 - 6 MB	NPU
hand_landmark_detector	ONNX	float	Snapdragon® X Elite	1.403 ms	6 - 6 MB	NPU
hand_landmark_detector	ONNX	float	Snapdragon® 8 Gen 3 Mobile	0.818 ms	0 - 59 MB	NPU
hand_landmark_detector	ONNX	float	Qualcomm® QCS8550 (Proxy)	1.108 ms	0 - 8 MB	NPU
hand_landmark_detector	ONNX	float	Qualcomm® QCS9075	1.811 ms	1 - 4 MB	NPU
hand_landmark_detector	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	0.658 ms	0 - 34 MB	NPU
hand_landmark_detector	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.517 ms	1 - 39 MB	NPU
hand_landmark_detector	QNN_DLC	float	Snapdragon® X2 Elite	0.867 ms	1 - 1 MB	NPU
hand_landmark_detector	QNN_DLC	float	Snapdragon® X Elite	1.299 ms	1 - 1 MB	NPU
hand_landmark_detector	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	0.765 ms	0 - 58 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	5.345 ms	1 - 35 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	1.026 ms	1 - 91 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® SA8775P	1.864 ms	1 - 39 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® QCS9075	1.696 ms	3 - 5 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	1.932 ms	0 - 53 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® SA7255P	5.345 ms	1 - 35 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® SA8295P	2.241 ms	0 - 30 MB	NPU
hand_landmark_detector	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	0.605 ms	1 - 39 MB	NPU
hand_landmark_detector	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.512 ms	0 - 44 MB	NPU
hand_landmark_detector	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	0.765 ms	0 - 61 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® QCS8275 (Proxy)	5.372 ms	0 - 41 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® QCS8550 (Proxy)	1.021 ms	0 - 3 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® SA8775P	1.882 ms	0 - 43 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® QCS9075	1.682 ms	0 - 9 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® QCS8450 (Proxy)	1.968 ms	0 - 57 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® SA7255P	5.372 ms	0 - 41 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® SA8295P	2.25 ms	1 - 36 MB	NPU
hand_landmark_detector	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	0.602 ms	0 - 43 MB	NPU

License

• The license for the original implementation of MediaPipe-Hand-Detection can be found here.

References

• MediaPipe Hands: On-device Real-time Hand Tracking
• 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.