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 2.45, ONNX Runtime 1.27.1 | Download |
| ONNX | w8a8 | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | Download |
| QNN_DLC | float | Universal | QAIRT 2.45 | Download |
| QNN_DLC | w8a8 | Universal | QAIRT 2.45 | Download |
| TFLITE | float | Universal | QAIRT 2.45 | Download |
| TFLITE | w8a8 | Universal | QAIRT 2.45 | 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:
- Model checkpoint: Imagenet
- Input resolution: 224x224
- Number of parameters: 1.37M
- Model size (float): 5.24 MB
- Model size (w8a8): 1.47 MB
Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |
|---|---|---|---|---|---|---|
| Shufflenet-v2 | ONNX | float | Snapdragon® X2 Elite | 0.435 ms | 2 - 2 MB | NPU |
| Shufflenet-v2 | ONNX | float | Snapdragon® X Elite | 0.841 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 0.514 ms | 0 - 40 MB | NPU |
| Shufflenet-v2 | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 0.952 ms | 1 - 44 MB | NPU |
| Shufflenet-v2 | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.821 ms | 0 - 5 MB | NPU |
| Shufflenet-v2 | ONNX | float | Qualcomm® QCS8450 | 0.952 ms | 1 - 44 MB | NPU |
| Shufflenet-v2 | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 0.97 ms | 1 - 3 MB | NPU |
| Shufflenet-v2 | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 0.841 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 0.446 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | ONNX | float | Snapdragon® 8 Elite Mobile | 0.446 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.439 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® X2 Elite | 0.346 ms | 1 - 1 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® X Elite | 0.578 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.403 ms | 0 - 40 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 0.645 ms | 0 - 44 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 0.935 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.578 ms | 0 - 4 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® QCS8450 | 0.645 ms | 0 - 44 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 0.692 ms | 0 - 3 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 0.578 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.352 ms | 0 - 34 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® 8 Elite Mobile | 0.352 ms | 0 - 34 MB | NPU |
| Shufflenet-v2 | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.345 ms | 0 - 30 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® X2 Elite | 0.389 ms | 1 - 1 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® X Elite | 0.922 ms | 1 - 1 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 0.513 ms | 0 - 37 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 1.353 ms | 0 - 43 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8275 | 1.716 ms | 1 - 26 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.785 ms | 1 - 23 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® SA8775P | 0.997 ms | 0 - 29 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® SA8650P | 0.997 ms | 0 - 29 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® SA8255P | 0.997 ms | 0 - 29 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® QCS8450 | 1.353 ms | 0 - 43 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 0.883 ms | 1 - 3 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 0.922 ms | 1 - 1 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 0.362 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® SA7255P | 1.716 ms | 1 - 26 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Qualcomm® SA8295P | 1.226 ms | 0 - 25 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 0.362 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.283 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 0.296 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® X Elite | 0.573 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.34 ms | 0 - 32 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 0.543 ms | 0 - 39 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 1.217 ms | 0 - 2 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8275 | 1.064 ms | 0 - 24 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.462 ms | 0 - 1 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® SA8775P | 0.637 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® SA8650P | 0.637 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® SA8255P | 0.637 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 0.543 ms | 0 - 39 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 0.546 ms | 2 - 4 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 0.573 ms | 0 - 0 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 1.337 ms | 0 - 23 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 0.459 ms | 0 - 23 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.262 ms | 0 - 23 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® SA7255P | 1.064 ms | 0 - 24 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Qualcomm® SA8295P | 0.846 ms | 0 - 22 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Mobile | 0.262 ms | 0 - 23 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.215 ms | 0 - 24 MB | NPU |
| Shufflenet-v2 | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 0.459 ms | 0 - 23 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 0.503 ms | 0 - 38 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 1.36 ms | 0 - 44 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® Dragonwing™ QCS8275 | 1.723 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.779 ms | 0 - 63 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® SA8775P | 1.024 ms | 0 - 29 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® SA8650P | 1.024 ms | 0 - 29 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® SA8255P | 1.024 ms | 0 - 29 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® QCS8450 | 1.36 ms | 0 - 44 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 1.025 ms | 0 - 5 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 0.37 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® SA7255P | 1.723 ms | 0 - 27 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Qualcomm® SA8295P | 1.231 ms | 0 - 23 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Snapdragon® 8 Elite Mobile | 0.37 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.285 ms | 0 - 28 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.321 ms | 0 - 32 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 0.512 ms | 0 - 37 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 0.858 ms | 0 - 4 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8275 | 1.031 ms | 0 - 25 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.457 ms | 0 - 4 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® SA8775P | 0.649 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® SA8650P | 0.649 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® SA8255P | 0.649 ms | 0 - 26 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® QCS8450 | 0.512 ms | 0 - 37 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 0.543 ms | 0 - 4 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 1.077 ms | 0 - 22 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 0.438 ms | 0 - 22 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.272 ms | 0 - 23 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® SA7255P | 1.031 ms | 0 - 25 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Qualcomm® SA8295P | 0.812 ms | 0 - 21 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Snapdragon® 8 Elite Mobile | 0.272 ms | 0 - 23 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.258 ms | 0 - 23 MB | NPU |
| Shufflenet-v2 | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 0.438 ms | 0 - 22 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.
