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 | 0.647 ms | 2 - 2 MB | NPU |
| hand_detector | ONNX | float | Snapdragon® X Elite | 0.851 ms | 0 - 0 MB | NPU |
| hand_detector | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 0.584 ms | 1 - 66 MB | NPU |
| hand_detector | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 1.491 ms | 0 - 63 MB | NPU |
| hand_detector | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.839 ms | 1 - 3 MB | NPU |
| hand_detector | ONNX | float | Qualcomm® QCS8450 | 1.491 ms | 0 - 63 MB | NPU |
| hand_detector | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 1.261 ms | 1 - 3 MB | NPU |
| hand_detector | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.434 ms | 0 - 46 MB | NPU |
| hand_detector | ONNX | float | Snapdragon® 8 Elite Mobile | 0.496 ms | 0 - 42 MB | NPU |
| hand_detector | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 0.496 ms | 0 - 42 MB | NPU |
| hand_detector | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 0.851 ms | 0 - 0 MB | NPU |
| hand_detector | QNN_DLC | float | Snapdragon® X2 Elite | 0.6 ms | 1 - 1 MB | NPU |
| hand_detector | QNN_DLC | float | Snapdragon® X Elite | 0.887 ms | 1 - 1 MB | NPU |
| hand_detector | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 0.537 ms | 0 - 57 MB | NPU |
| hand_detector | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 1.366 ms | 0 - 55 MB | NPU |
| hand_detector | QNN_DLC | float | Qualcomm® QCS8275 | 3.778 ms | 1 - 39 MB | NPU |
| hand_detector | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.728 ms | 1 - 2 MB | NPU |
| hand_detector | QNN_DLC | float | Qualcomm® SA8775P | 1.268 ms | 0 - 41 MB | NPU |
| hand_detector | QNN_DLC | float | Qualcomm® SA8650P | 1.268 ms | 0 - 41 MB | NPU |
| hand_detector | QNN_DLC | float | Qualcomm® SA8255P | 1.268 ms | 0 - 41 MB | NPU |
| hand_detector | QNN_DLC | float | Qualcomm® QCS8450 | 1.366 ms | 0 - 55 MB | NPU |
| hand_detector | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 1.116 ms | 1 - 3 MB | NPU |
| hand_detector | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.385 ms | 1 - 40 MB | NPU |
| hand_detector | QNN_DLC | float | Qualcomm® SA7255P | 3.778 ms | 1 - 39 MB | NPU |
| hand_detector | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 0.44 ms | 0 - 39 MB | NPU |
| hand_detector | QNN_DLC | float | Qualcomm® SA8295P | 1.696 ms | 0 - 31 MB | NPU |
| hand_detector | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 0.44 ms | 0 - 39 MB | NPU |
| hand_detector | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 0.887 ms | 1 - 1 MB | NPU |
| hand_detector | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 0.532 ms | 0 - 55 MB | NPU |
| hand_detector | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 1.38 ms | 0 - 54 MB | NPU |
| hand_detector | TFLITE | float | Qualcomm® QCS8275 | 3.822 ms | 0 - 37 MB | NPU |
| hand_detector | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.734 ms | 0 - 4 MB | NPU |
| hand_detector | TFLITE | float | Qualcomm® SA8775P | 1.301 ms | 0 - 41 MB | NPU |
| hand_detector | TFLITE | float | Qualcomm® SA8650P | 1.301 ms | 0 - 41 MB | NPU |
| hand_detector | TFLITE | float | Qualcomm® SA8255P | 1.301 ms | 0 - 41 MB | NPU |
| hand_detector | TFLITE | float | Qualcomm® QCS8450 | 1.38 ms | 0 - 54 MB | NPU |
| hand_detector | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 1.228 ms | 0 - 7 MB | NPU |
| hand_detector | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.389 ms | 0 - 36 MB | NPU |
| hand_detector | TFLITE | float | Qualcomm® SA7255P | 3.822 ms | 0 - 37 MB | NPU |
| hand_detector | TFLITE | float | Snapdragon® 8 Elite Mobile | 0.446 ms | 0 - 35 MB | NPU |
| hand_detector | TFLITE | float | Qualcomm® SA8295P | 1.717 ms | 0 - 30 MB | NPU |
| hand_detector | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 0.446 ms | 0 - 35 MB | NPU |
| hand_landmark_detector | ONNX | float | Snapdragon® X2 Elite | 0.711 ms | 2 - 2 MB | NPU |
| hand_landmark_detector | ONNX | float | Snapdragon® X Elite | 1.153 ms | 6 - 6 MB | NPU |
| hand_landmark_detector | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 0.823 ms | 0 - 61 MB | NPU |
| hand_landmark_detector | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 2.088 ms | 1 - 59 MB | NPU |
| hand_landmark_detector | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.092 ms | 0 - 3 MB | NPU |
| hand_landmark_detector | ONNX | float | Qualcomm® QCS8450 | 2.088 ms | 1 - 59 MB | NPU |
| hand_landmark_detector | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 1.831 ms | 1 - 3 MB | NPU |
| hand_landmark_detector | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.569 ms | 0 - 42 MB | NPU |
| hand_landmark_detector | ONNX | float | Snapdragon® 8 Elite Mobile | 0.66 ms | 0 - 38 MB | NPU |
| hand_landmark_detector | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 0.66 ms | 0 - 38 MB | NPU |
| hand_landmark_detector | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 1.153 ms | 6 - 6 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Snapdragon® X2 Elite | 0.842 ms | 1 - 1 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Snapdragon® X Elite | 1.253 ms | 1 - 1 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 0.77 ms | 0 - 57 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 1.945 ms | 0 - 54 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Qualcomm® QCS8275 | 5.306 ms | 1 - 35 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.997 ms | 1 - 79 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Qualcomm® SA8775P | 1.859 ms | 0 - 37 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Qualcomm® SA8650P | 1.859 ms | 0 - 37 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Qualcomm® SA8255P | 1.859 ms | 0 - 37 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Qualcomm® QCS8450 | 1.945 ms | 0 - 54 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 1.649 ms | 1 - 3 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.518 ms | 1 - 36 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Qualcomm® SA7255P | 5.306 ms | 1 - 35 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 0.6 ms | 1 - 34 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Qualcomm® SA8295P | 2.227 ms | 1 - 31 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 0.6 ms | 1 - 34 MB | NPU |
| hand_landmark_detector | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 1.253 ms | 1 - 1 MB | NPU |
| hand_landmark_detector | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 0.759 ms | 0 - 58 MB | NPU |
| hand_landmark_detector | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 1.946 ms | 0 - 61 MB | NPU |
| hand_landmark_detector | TFLITE | float | Qualcomm® QCS8275 | 5.387 ms | 0 - 39 MB | NPU |
| hand_landmark_detector | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.995 ms | 0 - 2 MB | NPU |
| hand_landmark_detector | TFLITE | float | Qualcomm® SA8775P | 1.891 ms | 0 - 43 MB | NPU |
| hand_landmark_detector | TFLITE | float | Qualcomm® SA8650P | 1.891 ms | 0 - 43 MB | NPU |
| hand_landmark_detector | TFLITE | float | Qualcomm® SA8255P | 1.891 ms | 0 - 43 MB | NPU |
| hand_landmark_detector | TFLITE | float | Qualcomm® QCS8450 | 1.946 ms | 0 - 61 MB | NPU |
| hand_landmark_detector | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 1.67 ms | 0 - 9 MB | NPU |
| hand_landmark_detector | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.516 ms | 0 - 38 MB | NPU |
| hand_landmark_detector | TFLITE | float | Qualcomm® SA7255P | 5.387 ms | 0 - 39 MB | NPU |
| hand_landmark_detector | TFLITE | float | Snapdragon® 8 Elite Mobile | 0.605 ms | 0 - 36 MB | NPU |
| hand_landmark_detector | TFLITE | float | Qualcomm® SA8295P | 2.271 ms | 0 - 34 MB | NPU |
| hand_landmark_detector | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 0.605 ms | 0 - 36 MB | NPU |
License
- The license for the original implementation of MediaPipe-Hand-Detection can be found here.
References
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.
