MobileNet-v3-Small: Optimized for Qualcomm Devices
MobileNetV3Small 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 MobileNet-v3-Small 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 1.30.0 | Download |
| ONNX | w8a16_mixed_int16 | Universal | QAIRT 2.50, ONNX Runtime 1.30.0 | Download |
| QNN_DLC | float | Universal | QAIRT 2.50 | Download |
| QNN_DLC | w8a16 | Universal | QAIRT 2.45 | Download |
| QNN_DLC | w8a16_mixed_int16 | Universal | QAIRT 2.50 | Download |
| TFLITE | float | Universal | QAIRT 2.50 | Download |
For more device-specific assets and performance metrics, visit MobileNet-v3-Small 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 MobileNet-v3-Small 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): 9.71 MB
- Number of parameters: 2.54M
Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |
|---|---|---|---|---|---|---|
| MobileNet-v3-Small | ONNX | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.248 ms | 0 - 34 MB | NPU |
| MobileNet-v3-Small | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 0.286 ms | 0 - 37 MB | NPU |
| MobileNet-v3-Small | ONNX | float | Snapdragon® X2 Elite | 0.248 ms | 2 - 2 MB | NPU |
| MobileNet-v3-Small | ONNX | float | Snapdragon® X Elite | 0.554 ms | 5 - 5 MB | NPU |
| MobileNet-v3-Small | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 0.364 ms | 0 - 49 MB | NPU |
| MobileNet-v3-Small | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 0.802 ms | 1 - 57 MB | NPU |
| MobileNet-v3-Small | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 0.663 ms | 1 - 5 MB | NPU |
| MobileNet-v3-Small | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.552 ms | 0 - 9 MB | NPU |
| MobileNet-v3-Small | ONNX | float | Qualcomm® QCS8450 | 0.802 ms | 1 - 57 MB | NPU |
| MobileNet-v3-Small | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 0.72 ms | 1 - 4 MB | NPU |
| MobileNet-v3-Small | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 0.554 ms | 5 - 5 MB | NPU |
| MobileNet-v3-Small | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 0.286 ms | 0 - 37 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.241 ms | 0 - 37 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Snapdragon® 8 Elite For Galaxy Mobile | 0.279 ms | 0 - 36 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Snapdragon® X2 Elite | 0.235 ms | 1 - 1 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Snapdragon® X Elite | 0.568 ms | 0 - 0 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Snapdragon® 8 Gen 3 Mobile | 0.383 ms | 0 - 51 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Qualcomm® Dragonwing™ IQ-8275 | 0.601 ms | 0 - 4 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.571 ms | 0 - 6 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Qualcomm® Dragonwing™ IQ-9075 | 0.702 ms | 0 - 3 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Qualcomm® Dragonwing™ IQ-X7181 | 0.568 ms | 0 - 0 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Qualcomm® Dragonwing™ Q-6690 | 2.09 ms | 0 - 38 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Qualcomm® Dragonwing™ Q-7790 | 0.591 ms | 0 - 37 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Qualcomm® Dragonwing™ Q-8750 | 0.279 ms | 0 - 36 MB | NPU |
| MobileNet-v3-Small | ONNX | w8a16_mixed_int16 | Snapdragon® 7 Gen 4 Mobile | 0.591 ms | 0 - 37 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.32 ms | 1 - 35 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 0.416 ms | 0 - 36 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Snapdragon® X2 Elite | 0.436 ms | 1 - 1 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Snapdragon® X Elite | 0.986 ms | 1 - 1 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 0.559 ms | 0 - 47 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 1.466 ms | 0 - 52 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 0.9 ms | 1 - 4 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.838 ms | 1 - 2 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Qualcomm® SA8775P | 1.133 ms | 1 - 35 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Qualcomm® SA8650P | 1.133 ms | 1 - 35 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Qualcomm® SA8255P | 1.133 ms | 1 - 35 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Qualcomm® QCS8450 | 1.466 ms | 0 - 52 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 1.029 ms | 3 - 5 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 0.986 ms | 1 - 1 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 0.416 ms | 0 - 36 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Qualcomm® SA7255P | 2.094 ms | 1 - 33 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | float | Qualcomm® SA8295P | 1.345 ms | 1 - 33 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.311 ms | 0 - 28 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 0.376 ms | 0 - 31 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 0.426 ms | 0 - 0 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Snapdragon® X Elite | 0.941 ms | 1 - 1 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 0.545 ms | 0 - 39 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Snapdragon® 8 Gen 1 Mobile | 0.99 ms | 0 - 47 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS6490 | 2.669 ms | 2 - 4 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-8275 | 0.812 ms | 0 - 3 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.803 ms | 0 - 2 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® SA8775P | 1.009 ms | 0 - 30 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® SA8650P | 1.009 ms | 0 - 30 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® SA8255P | 1.009 ms | 0 - 30 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® QCS8450 | 0.99 ms | 0 - 47 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 0.901 ms | 0 - 2 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 0.941 ms | 1 - 1 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-6690 | 2.833 ms | 0 - 142 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-7790 | 0.808 ms | 0 - 26 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 0.376 ms | 0 - 31 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® SA7255P | 1.725 ms | 0 - 28 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Qualcomm® SA8295P | 1.327 ms | 0 - 27 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 0.808 ms | 0 - 26 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.323 ms | 0 - 33 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Snapdragon® 8 Elite For Galaxy Mobile | 0.397 ms | 0 - 35 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Snapdragon® X2 Elite | 0.42 ms | 0 - 0 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Snapdragon® X Elite | 0.969 ms | 0 - 0 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Snapdragon® 8 Gen 3 Mobile | 0.577 ms | 0 - 43 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Qualcomm® Dragonwing™ IQ-8275 | 0.827 ms | 0 - 3 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.843 ms | 0 - 2 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Qualcomm® SA8650P | 1.02 ms | 0 - 30 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Qualcomm® SA8255P | 1.02 ms | 0 - 30 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Qualcomm® Dragonwing™ IQ-9075 | 0.934 ms | 2 - 5 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Qualcomm® Dragonwing™ IQ-X7181 | 0.969 ms | 0 - 0 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Qualcomm® Dragonwing™ Q-6690 | 2.88 ms | 0 - 138 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Qualcomm® Dragonwing™ Q-7790 | 0.835 ms | 0 - 31 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Qualcomm® Dragonwing™ Q-8750 | 0.397 ms | 0 - 35 MB | NPU |
| MobileNet-v3-Small | QNN_DLC | w8a16_mixed_int16 | Snapdragon® 7 Gen 4 Mobile | 0.835 ms | 0 - 31 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.321 ms | 0 - 35 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 0.43 ms | 0 - 32 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 0.57 ms | 0 - 49 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 1.477 ms | 0 - 56 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 0.929 ms | 0 - 10 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.857 ms | 0 - 2 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Qualcomm® SA8775P | 1.158 ms | 0 - 37 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Qualcomm® SA8650P | 1.158 ms | 0 - 37 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Qualcomm® SA8255P | 1.158 ms | 0 - 37 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Qualcomm® QCS8450 | 1.477 ms | 0 - 56 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 1.027 ms | 0 - 9 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 0.43 ms | 0 - 32 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Qualcomm® SA7255P | 2.147 ms | 0 - 33 MB | NPU |
| MobileNet-v3-Small | TFLITE | float | Qualcomm® SA8295P | 1.369 ms | 0 - 33 MB | NPU |
License
- The license for the original implementation of MobileNet-v3-Small 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.
