Shufflenet-v2: Optimized for Qualcomm Devices
ShufflenetV2 is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.
This is based on the implementation of Shufflenet-v2 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 2.50, ONNX Runtime 1.27.1 | Download |
| ONNX | w8a16 | Universal | QAIRT 2.50, ONNX Runtime 1.27.1 | Download |
| ONNX | w8a8 | Universal | QAIRT 2.50, ONNX Runtime 1.27.1 | Download |
| QNN_DLC | float | Universal | QAIRT 2.50 | Download |
| QNN_DLC | w8a16 | Universal | QAIRT 2.50 | Download |
| QNN_DLC | w8a8 | Universal | QAIRT 2.50 | Download |
| TFLITE | float | Universal | QAIRT 2.50 | Download |
| TFLITE | w8a8 | Universal | QAIRT 2.50 | Download |
For more device-specific assets and performance metrics, visit Shufflenet-v2 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 Shufflenet-v2 on GitHub for usage instructions.
Model Details
Model Type: Model_use_case.image_classification
Model Stats:
- Input resolution: 224x224
- Model checkpoint: Imagenet
- Model size (float): 5.24 MB
- Model size (w8a8): 1.47 MB
- Number of parameters: 1.37M
Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |
|---|---|---|---|---|---|---|
| Shufflenet-v2 | ONNX | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.407 ms | 0 - 31 MB | NPU |
| Shufflenet-v2 | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 0.441 ms | 0 - 30 MB | NPU |
| Shufflenet-v2 | ONNX | float | Snapdragon® X2 Elite | 0.404 ms | 2 - 2 MB | NPU |
| Shufflenet-v2 | ONNX | float | Snapdragon® X Elite | 0.842 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 0.529 ms | 0 - 44 MB | NPU |
| Shufflenet-v2 | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 0.949 ms | 1 - 50 MB | NPU |
| Shufflenet-v2 | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 0.833 ms | 0 - 4 MB | NPU |
| Shufflenet-v2 | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.82 ms | 0 - 5 MB | NPU |
| Shufflenet-v2 | ONNX | float | Qualcomm® QCS8450 | 0.949 ms | 1 - 50 MB | NPU |
| Shufflenet-v2 | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 0.994 ms | 1 - 4 MB | NPU |
| Shufflenet-v2 | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 0.842 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 0.441 ms | 0 - 30 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.396 ms | 0 - 35 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 0.418 ms | 0 - 38 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Snapdragon® X2 Elite | 0.386 ms | 1 - 1 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Snapdragon® X Elite | 0.827 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 0.527 ms | 0 - 45 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Snapdragon® 8 Gen 1 Mobile | 0.88 ms | 0 - 51 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS6490 | 1.524 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-8275 | 0.783 ms | 0 - 4 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.81 ms | 0 - 4 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Qualcomm® QCS8450 | 0.88 ms | 0 - 51 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 0.923 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 0.827 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-6690 | 1.99 ms | 0 - 34 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-7790 | 0.805 ms | 0 - 34 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 0.418 ms | 0 - 38 MB | NPU |
| Shufflenet-v2 | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 0.805 ms | 0 - 34 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.335 ms | 0 - 35 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.353 ms | 0 - 33 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® X2 Elite | 0.331 ms | 1 - 1 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® X Elite | 0.58 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.408 ms | 0 - 43 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 0.617 ms | 0 - 48 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 1.057 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 0.611 ms | 0 - 4 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.575 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® QCS8450 | 0.617 ms | 0 - 48 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 0.693 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 0.58 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 1.157 ms | 0 - 34 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 0.467 ms | 0 - 33 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.353 ms | 0 - 33 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 0.467 ms | 0 - 33 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.28 ms | 0 - 32 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 0.373 ms | 1 - 30 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® X2 Elite | 0.371 ms | 1 - 1 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® X Elite | 0.902 ms | 1 - 1 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 0.505 ms | 0 - 42 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 1.307 ms | 0 - 43 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 0.775 ms | 1 - 4 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.775 ms | 1 - 2 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® SA8775P | 1.009 ms | 1 - 32 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® SA8650P | 1.009 ms | 1 - 32 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® SA8255P | 1.009 ms | 1 - 32 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® QCS8450 | 1.307 ms | 0 - 43 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 0.897 ms | 3 - 5 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 0.902 ms | 1 - 1 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 0.373 ms | 1 - 30 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® SA8295P | 1.13 ms | 0 - 28 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.282 ms | 0 - 28 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 0.338 ms | 0 - 33 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 0.404 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Snapdragon® X Elite | 0.865 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 0.518 ms | 0 - 37 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Snapdragon® 8 Gen 1 Mobile | 0.817 ms | 0 - 44 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS6490 | 1.798 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-8275 | 0.762 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.745 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® SA8650P | 0.922 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® SA8255P | 0.922 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® QCS8450 | 0.817 ms | 0 - 44 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 0.854 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 0.865 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-6690 | 2.211 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-7790 | 0.723 ms | 0 - 28 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 0.338 ms | 0 - 33 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Qualcomm® SA8295P | 1.004 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 0.723 ms | 0 - 28 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.213 ms | 0 - 28 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.263 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 0.306 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® X Elite | 0.593 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.34 ms | 0 - 37 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 0.52 ms | 0 - 40 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 1.16 ms | 0 - 2 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 0.51 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.469 ms | 0 - 2 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 0.52 ms | 0 - 40 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 0.548 ms | 0 - 2 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 0.593 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 1.335 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 0.462 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.263 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® SA8295P | 0.726 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 0.462 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.294 ms | 0 - 31 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 0.371 ms | 0 - 29 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 0.508 ms | 0 - 42 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 1.33 ms | 0 - 47 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 0.814 ms | 0 - 7 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.785 ms | 0 - 2 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® SA8775P | 1.02 ms | 0 - 33 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® SA8650P | 1.02 ms | 0 - 33 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® SA8255P | 1.02 ms | 0 - 33 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® QCS8450 | 1.33 ms | 0 - 47 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 0.911 ms | 0 - 6 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 0.371 ms | 0 - 29 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® SA7255P | 1.778 ms | 0 - 30 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® SA8295P | 1.115 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.256 ms | 0 - 29 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.271 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.321 ms | 0 - 37 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 0.499 ms | 0 - 43 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 0.792 ms | 0 - 4 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 0.504 ms | 0 - 5 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.456 ms | 0 - 7 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® SA8775P | 0.639 ms | 0 - 30 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® SA8650P | 0.639 ms | 0 - 30 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® SA8255P | 0.639 ms | 0 - 30 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® QCS8450 | 0.499 ms | 0 - 43 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 0.556 ms | 0 - 4 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 1.081 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 0.449 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.271 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® SA7255P | 1.056 ms | 0 - 29 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® SA8295P | 0.684 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 0.449 ms | 0 - 27 MB | NPU |
License
- The license for the original implementation of Shufflenet-v2 can be found here.
References
- ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design
- 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.
