ConvNext-Tiny: Optimized for Qualcomm Devices
ConvNextTiny 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 ConvNext-Tiny 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 |
| ONNX | w8a16 | Universal | QAIRT 2.45, ONNX Runtime 1.25.0 | Download |
| QNN_DLC | float | Universal | QAIRT 2.45 | Download |
| QNN_DLC | w8a16 | Universal | QAIRT 2.45 | Download |
| TFLITE | float | Universal | QAIRT 2.45 | Download |
For more device-specific assets and performance metrics, visit ConvNext-Tiny 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 ConvNext-Tiny 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: 28.6M
- Model size (float): 109 MB
- Model size (w8a16): 28.9 MB
Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |
|---|---|---|---|---|---|---|
| ConvNext-Tiny | ONNX | float | Snapdragon® X2 Elite | 1.336 ms | 2 - 2 MB | NPU |
| ConvNext-Tiny | ONNX | float | Snapdragon® X Elite | 2.694 ms | 57 - 57 MB | NPU |
| ConvNext-Tiny | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 2.03 ms | 0 - 126 MB | NPU |
| ConvNext-Tiny | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 9.047 ms | 1 - 130 MB | NPU |
| ConvNext-Tiny | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.724 ms | 0 - 130 MB | NPU |
| ConvNext-Tiny | ONNX | float | Qualcomm® QCS8450 | 9.047 ms | 1 - 130 MB | NPU |
| ConvNext-Tiny | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.296 ms | 0 - 73 MB | NPU |
| ConvNext-Tiny | ONNX | float | Snapdragon® 8 Elite Mobile | 1.574 ms | 0 - 75 MB | NPU |
| ConvNext-Tiny | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 3.849 ms | 1 - 4 MB | NPU |
| ConvNext-Tiny | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 1.574 ms | 0 - 75 MB | NPU |
| ConvNext-Tiny | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 2.694 ms | 57 - 57 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Snapdragon® X2 Elite | 0.998 ms | 1 - 1 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Snapdragon® X Elite | 2.227 ms | 30 - 30 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 1.545 ms | 0 - 133 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.165 ms | 0 - 6 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Snapdragon® 8 Elite Mobile | 1.146 ms | 0 - 109 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 2.337 ms | 0 - 3 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 2.592 ms | 0 - 118 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 0.94 ms | 0 - 109 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-6690 | 21.741 ms | 0 - 239 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-7790 | 2.592 ms | 0 - 118 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 1.146 ms | 0 - 109 MB | NPU |
| ConvNext-Tiny | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 2.227 ms | 30 - 30 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Snapdragon® X2 Elite | 0.631 ms | 1 - 1 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Snapdragon® X Elite | 1.377 ms | 29 - 29 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.97 ms | 0 - 114 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 2.123 ms | 0 - 116 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 5.661 ms | 0 - 3 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.392 ms | 0 - 21 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Qualcomm® QCS8450 | 2.123 ms | 0 - 116 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Snapdragon® 8 Elite Mobile | 0.781 ms | 0 - 81 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.734 ms | 0 - 81 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 1.508 ms | 0 - 3 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.718 ms | 0 - 88 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 10.696 ms | 0 - 87 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 1.734 ms | 0 - 81 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.781 ms | 0 - 81 MB | NPU |
| ConvNext-Tiny | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 1.377 ms | 29 - 29 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Snapdragon® X2 Elite | 1.996 ms | 1 - 1 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Snapdragon® X Elite | 3.692 ms | 1 - 1 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 2.483 ms | 0 - 126 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 9.48 ms | 0 - 129 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Qualcomm® QCS8275 | 14.83 ms | 1 - 76 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 3.428 ms | 1 - 2 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Qualcomm® SA8775P | 4.781 ms | 1 - 78 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Qualcomm® SA8650P | 4.781 ms | 1 - 78 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Qualcomm® SA8255P | 4.781 ms | 1 - 78 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Qualcomm® QCS8450 | 9.48 ms | 0 - 129 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.584 ms | 1 - 80 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 1.916 ms | 0 - 77 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 5.332 ms | 3 - 5 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Qualcomm® SA7255P | 14.83 ms | 1 - 76 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Qualcomm® SA8295P | 8.739 ms | 1 - 79 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 1.916 ms | 0 - 77 MB | NPU |
| ConvNext-Tiny | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 3.692 ms | 1 - 1 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 1.652 ms | 0 - 0 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Snapdragon® X Elite | 3.376 ms | 0 - 0 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 2.15 ms | 0 - 124 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Qualcomm® QCS8275 | 6.843 ms | 0 - 98 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 3.1 ms | 0 - 2 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Qualcomm® SA8775P | 3.524 ms | 0 - 100 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Qualcomm® SA8650P | 3.524 ms | 0 - 100 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Qualcomm® SA8255P | 3.524 ms | 0 - 100 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Snapdragon® 8 Elite Mobile | 1.598 ms | 0 - 100 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 3.309 ms | 0 - 2 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Qualcomm® SA7255P | 6.843 ms | 0 - 98 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 3.424 ms | 0 - 111 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 1.297 ms | 0 - 101 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-6690 | 23.469 ms | 0 - 251 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-7790 | 3.424 ms | 0 - 111 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 1.598 ms | 0 - 100 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 3.376 ms | 0 - 0 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 0.942 ms | 0 - 0 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Snapdragon® X Elite | 1.947 ms | 0 - 0 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.241 ms | 0 - 108 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 2.589 ms | 0 - 106 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 6.532 ms | 0 - 2 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® QCS8275 | 3.82 ms | 0 - 70 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.807 ms | 0 - 113 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® SA8775P | 2.045 ms | 0 - 70 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® SA8650P | 2.045 ms | 0 - 70 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® SA8255P | 2.045 ms | 0 - 70 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 2.589 ms | 0 - 106 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Snapdragon® 8 Elite Mobile | 0.974 ms | 0 - 73 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 2.146 ms | 0 - 73 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 1.909 ms | 0 - 2 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® SA7255P | 3.82 ms | 0 - 70 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® SA8295P | 3.007 ms | 0 - 67 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.847 ms | 0 - 81 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 12.049 ms | 0 - 96 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 2.146 ms | 0 - 73 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.974 ms | 0 - 73 MB | NPU |
| ConvNext-Tiny | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 1.947 ms | 0 - 0 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 2.135 ms | 0 - 127 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 8.897 ms | 0 - 124 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Qualcomm® QCS8275 | 14.03 ms | 0 - 73 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.816 ms | 0 - 3 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Qualcomm® SA8775P | 4.27 ms | 0 - 75 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Qualcomm® SA8650P | 4.27 ms | 0 - 75 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Qualcomm® SA8255P | 4.27 ms | 0 - 75 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Qualcomm® QCS8450 | 8.897 ms | 0 - 124 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.335 ms | 0 - 74 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Snapdragon® 8 Elite Mobile | 1.586 ms | 0 - 76 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 4.05 ms | 0 - 59 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Qualcomm® SA7255P | 14.03 ms | 0 - 73 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Qualcomm® SA8295P | 7.856 ms | 0 - 72 MB | NPU |
| ConvNext-Tiny | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 1.586 ms | 0 - 76 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.207 ms | 0 - 111 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 2.663 ms | 0 - 111 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 42.975 ms | 3 - 56 MB | CPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® QCS8275 | 3.757 ms | 0 - 75 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.744 ms | 0 - 47 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® SA8775P | 2.008 ms | 0 - 76 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® SA8650P | 2.008 ms | 0 - 76 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® SA8255P | 2.008 ms | 0 - 76 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® QCS8450 | 2.663 ms | 0 - 111 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Snapdragon® 8 Elite Mobile | 0.91 ms | 0 - 74 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 11.549 ms | 2 - 44 MB | CPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 1.858 ms | 0 - 31 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® SA7255P | 3.757 ms | 0 - 75 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® SA8295P | 2.996 ms | 0 - 70 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.786 ms | 0 - 82 MB | NPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 27.008 ms | 4 - 47 MB | CPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 11.549 ms | 2 - 44 MB | CPU |
| ConvNext-Tiny | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.91 ms | 0 - 74 MB | NPU |
License
- The license for the original implementation of ConvNext-Tiny 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.
