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 1.24.1	Download
QNN_DLC	float	Universal	QAIRT 2.43	Download
TFLITE	float	Universal	QAIRT 2.43, TFLite 2.17.0	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 (HandDetector): 1.76M
• Model size (HandDetector) (float): 6.75 MB
• Number of parameters (HandLandmarkDetector): 2.01M
• Model size (HandLandmarkDetector) (float): 7.70 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
HandDetector	ONNX	float	Snapdragon® X2 Elite	0.487 ms	0 - 0 MB	NPU
HandDetector	ONNX	float	Snapdragon® X Elite	1.014 ms	3 - 3 MB	NPU
HandDetector	ONNX	float	Snapdragon® 8 Gen 3 Mobile	0.598 ms	0 - 65 MB	NPU
HandDetector	ONNX	float	Qualcomm® QCS8550 (Proxy)	0.839 ms	0 - 14 MB	NPU
HandDetector	ONNX	float	Qualcomm® QCS9075	1.275 ms	1 - 3 MB	NPU
HandDetector	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	0.495 ms	0 - 38 MB	NPU
HandDetector	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.419 ms	0 - 47 MB	NPU
HandDetector	QNN_DLC	float	Snapdragon® X2 Elite	0.617 ms	1 - 1 MB	NPU
HandDetector	QNN_DLC	float	Snapdragon® X Elite	0.908 ms	1 - 1 MB	NPU
HandDetector	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	0.534 ms	0 - 59 MB	NPU
HandDetector	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	3.782 ms	1 - 42 MB	NPU
HandDetector	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	0.733 ms	0 - 10 MB	NPU
HandDetector	QNN_DLC	float	Qualcomm® SA8775P	1.297 ms	1 - 43 MB	NPU
HandDetector	QNN_DLC	float	Qualcomm® QCS9075	1.13 ms	1 - 3 MB	NPU
HandDetector	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	1.349 ms	0 - 53 MB	NPU
HandDetector	QNN_DLC	float	Qualcomm® SA7255P	3.782 ms	1 - 42 MB	NPU
HandDetector	QNN_DLC	float	Qualcomm® SA8295P	1.673 ms	0 - 31 MB	NPU
HandDetector	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	0.445 ms	0 - 43 MB	NPU
HandDetector	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.381 ms	1 - 44 MB	NPU
HandDetector	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	0.536 ms	0 - 58 MB	NPU
HandDetector	TFLITE	float	Qualcomm® QCS8275 (Proxy)	3.836 ms	0 - 38 MB	NPU
HandDetector	TFLITE	float	Qualcomm® QCS8550 (Proxy)	0.74 ms	0 - 4 MB	NPU
HandDetector	TFLITE	float	Qualcomm® SA8775P	1.338 ms	0 - 42 MB	NPU
HandDetector	TFLITE	float	Qualcomm® QCS9075	1.138 ms	0 - 7 MB	NPU
HandDetector	TFLITE	float	Qualcomm® QCS8450 (Proxy)	1.368 ms	0 - 50 MB	NPU
HandDetector	TFLITE	float	Qualcomm® SA7255P	3.836 ms	0 - 38 MB	NPU
HandDetector	TFLITE	float	Qualcomm® SA8295P	1.714 ms	0 - 31 MB	NPU
HandDetector	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	0.452 ms	0 - 36 MB	NPU
HandDetector	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.379 ms	0 - 42 MB	NPU
HandLandmarkDetector	ONNX	float	Snapdragon® X2 Elite	0.719 ms	6 - 6 MB	NPU
HandLandmarkDetector	ONNX	float	Snapdragon® X Elite	1.386 ms	6 - 6 MB	NPU
HandLandmarkDetector	ONNX	float	Snapdragon® 8 Gen 3 Mobile	0.832 ms	0 - 69 MB	NPU
HandLandmarkDetector	ONNX	float	Qualcomm® QCS8550 (Proxy)	1.1 ms	0 - 11 MB	NPU
HandLandmarkDetector	ONNX	float	Qualcomm® QCS9075	1.83 ms	1 - 4 MB	NPU
HandLandmarkDetector	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	0.661 ms	0 - 45 MB	NPU
HandLandmarkDetector	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.559 ms	0 - 43 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Snapdragon® X2 Elite	0.876 ms	1 - 1 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Snapdragon® X Elite	1.283 ms	1 - 1 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	0.754 ms	0 - 57 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	5.336 ms	1 - 35 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	1.005 ms	0 - 7 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Qualcomm® SA8775P	1.893 ms	1 - 39 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Qualcomm® QCS9075	1.692 ms	3 - 5 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	1.947 ms	0 - 52 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Qualcomm® SA7255P	5.336 ms	1 - 35 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Qualcomm® SA8295P	2.22 ms	0 - 30 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	0.601 ms	0 - 34 MB	NPU
HandLandmarkDetector	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.513 ms	1 - 39 MB	NPU
HandLandmarkDetector	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	0.764 ms	0 - 60 MB	NPU
HandLandmarkDetector	TFLITE	float	Qualcomm® QCS8275 (Proxy)	5.351 ms	0 - 41 MB	NPU
HandLandmarkDetector	TFLITE	float	Qualcomm® QCS8550 (Proxy)	1.028 ms	0 - 2 MB	NPU
HandLandmarkDetector	TFLITE	float	Qualcomm® SA8775P	1.89 ms	0 - 44 MB	NPU
HandLandmarkDetector	TFLITE	float	Qualcomm® QCS9075	1.679 ms	0 - 9 MB	NPU
HandLandmarkDetector	TFLITE	float	Qualcomm® QCS8450 (Proxy)	1.97 ms	0 - 57 MB	NPU
HandLandmarkDetector	TFLITE	float	Qualcomm® SA7255P	5.351 ms	0 - 41 MB	NPU
HandLandmarkDetector	TFLITE	float	Qualcomm® SA8295P	2.232 ms	0 - 35 MB	NPU
HandLandmarkDetector	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	0.605 ms	0 - 43 MB	NPU
HandLandmarkDetector	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.51 ms	0 - 44 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.