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 2.45, ONNX Runtime 1.25.0	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 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® X2 Elite	0.485 ms	212 - 212 MB	NPU
hand_detector	ONNX	float	Snapdragon® X Elite	0.854 ms	177 - 177 MB	NPU
hand_detector	ONNX	float	Snapdragon® 8 Gen 3 Mobile	0.589 ms	0 - 61 MB	NPU
hand_detector	ONNX	float	Snapdragon® 8 Gen 1 Mobile	1.499 ms	1 - 54 MB	NPU
hand_detector	ONNX	float	Qualcomm® QCS8550 (Proxy)	0.854 ms	0 - 72 MB	NPU
hand_detector	ONNX	float	Qualcomm® QCS8450	1.499 ms	1 - 54 MB	NPU
hand_detector	ONNX	float	Snapdragon® 8 Elite Mobile	0.503 ms	0 - 41 MB	NPU
hand_detector	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.425 ms	1 - 42 MB	NPU
hand_detector	ONNX	float	Qualcomm® QCS9075	1.263 ms	0 - 49 MB	NPU
hand_detector	ONNX	float	Qualcomm® QCS8750	0.503 ms	0 - 41 MB	NPU
hand_detector	ONNX	float	Qualcomm® QCS7181	0.854 ms	177 - 177 MB	NPU
hand_detector	QNN_DLC	float	Snapdragon® X2 Elite	0.617 ms	1 - 1 MB	NPU
hand_detector	QNN_DLC	float	Snapdragon® X Elite	0.902 ms	1 - 1 MB	NPU
hand_detector	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	0.53 ms	0 - 57 MB	NPU
hand_detector	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	1.368 ms	0 - 56 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® QCS8275	3.779 ms	1 - 40 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	0.734 ms	1 - 41 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® QCS8450	1.368 ms	0 - 56 MB	NPU
hand_detector	QNN_DLC	float	Snapdragon® 8 Elite Mobile	0.442 ms	0 - 43 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® SA8295P	1.701 ms	0 - 31 MB	NPU
hand_detector	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.379 ms	1 - 39 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® SA7255P	3.779 ms	1 - 40 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® QCS9075	1.115 ms	3 - 5 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® QCS8750	0.442 ms	0 - 43 MB	NPU
hand_detector	QNN_DLC	float	Qualcomm® QCS7181	0.902 ms	1 - 1 MB	NPU
hand_detector	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	0.534 ms	0 - 56 MB	NPU
hand_detector	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	1.379 ms	0 - 55 MB	NPU
hand_detector	TFLITE	float	Qualcomm® QCS8275	3.835 ms	0 - 37 MB	NPU
hand_detector	TFLITE	float	Qualcomm® QCS8550 (Proxy)	0.727 ms	0 - 2 MB	NPU
hand_detector	TFLITE	float	Qualcomm® SA8775P	7.407 ms	0 - 34 MB	GPU
hand_detector	TFLITE	float	Qualcomm® SA8650P	7.407 ms	0 - 34 MB	GPU
hand_detector	TFLITE	float	Qualcomm® SA8255P	7.407 ms	0 - 34 MB	GPU
hand_detector	TFLITE	float	Qualcomm® QCS8450	1.379 ms	0 - 55 MB	NPU
hand_detector	TFLITE	float	Snapdragon® 8 Elite Mobile	0.446 ms	0 - 42 MB	NPU
hand_detector	TFLITE	float	Qualcomm® SA8295P	1.735 ms	0 - 30 MB	NPU
hand_detector	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.38 ms	0 - 36 MB	NPU
hand_detector	TFLITE	float	Qualcomm® SA7255P	3.835 ms	0 - 37 MB	NPU
hand_detector	TFLITE	float	Qualcomm® QCS9075	1.129 ms	0 - 7 MB	NPU
hand_detector	TFLITE	float	Qualcomm® QCS8750	0.446 ms	0 - 42 MB	NPU
hand_landmark_detector	ONNX	float	Snapdragon® X2 Elite	0.714 ms	212 - 212 MB	NPU
hand_landmark_detector	ONNX	float	Snapdragon® X Elite	1.164 ms	148 - 148 MB	NPU
hand_landmark_detector	ONNX	float	Snapdragon® 8 Gen 3 Mobile	0.809 ms	0 - 49 MB	NPU
hand_landmark_detector	ONNX	float	Snapdragon® 8 Gen 1 Mobile	2.074 ms	1 - 51 MB	NPU
hand_landmark_detector	ONNX	float	Qualcomm® QCS8550 (Proxy)	1.187 ms	0 - 14 MB	NPU
hand_landmark_detector	ONNX	float	Qualcomm® QCS8450	2.074 ms	1 - 51 MB	NPU
hand_landmark_detector	ONNX	float	Snapdragon® 8 Elite Mobile	0.658 ms	0 - 37 MB	NPU
hand_landmark_detector	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.56 ms	0 - 35 MB	NPU
hand_landmark_detector	ONNX	float	Qualcomm® QCS9075	1.84 ms	0 - 51 MB	NPU
hand_landmark_detector	ONNX	float	Qualcomm® QCS8750	0.658 ms	0 - 37 MB	NPU
hand_landmark_detector	ONNX	float	Qualcomm® QCS7181	1.164 ms	148 - 148 MB	NPU
hand_landmark_detector	QNN_DLC	float	Snapdragon® X2 Elite	0.877 ms	1 - 1 MB	NPU
hand_landmark_detector	QNN_DLC	float	Snapdragon® X Elite	1.295 ms	1 - 1 MB	NPU
hand_landmark_detector	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	0.761 ms	0 - 56 MB	NPU
hand_landmark_detector	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	1.937 ms	0 - 52 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® QCS8275	5.318 ms	1 - 36 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	1.083 ms	1 - 14 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® QCS8450	1.937 ms	0 - 52 MB	NPU
hand_landmark_detector	QNN_DLC	float	Snapdragon® 8 Elite Mobile	0.6 ms	0 - 38 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® SA8295P	2.24 ms	0 - 30 MB	NPU
hand_landmark_detector	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.517 ms	1 - 35 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® SA7255P	5.318 ms	1 - 36 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® QCS9075	1.671 ms	1 - 3 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® QCS8750	0.6 ms	0 - 38 MB	NPU
hand_landmark_detector	QNN_DLC	float	Qualcomm® QCS7181	1.295 ms	1 - 1 MB	NPU
hand_landmark_detector	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	0.759 ms	0 - 59 MB	NPU
hand_landmark_detector	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	1.938 ms	0 - 61 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® QCS8275	5.346 ms	0 - 39 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® QCS8550 (Proxy)	1.085 ms	0 - 3 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® SA8775P	4.771 ms	0 - 35 MB	GPU
hand_landmark_detector	TFLITE	float	Qualcomm® SA8650P	4.771 ms	0 - 35 MB	GPU
hand_landmark_detector	TFLITE	float	Qualcomm® SA8255P	4.771 ms	0 - 35 MB	GPU
hand_landmark_detector	TFLITE	float	Qualcomm® QCS8450	1.938 ms	0 - 61 MB	NPU
hand_landmark_detector	TFLITE	float	Snapdragon® 8 Elite Mobile	0.609 ms	0 - 43 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® SA8295P	2.254 ms	0 - 34 MB	NPU
hand_landmark_detector	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.521 ms	0 - 38 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® SA7255P	5.346 ms	0 - 39 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® QCS9075	1.683 ms	0 - 9 MB	NPU
hand_landmark_detector	TFLITE	float	Qualcomm® QCS8750	0.609 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.