NAFSSR: Stereo Image Super-Resolution Using NAFNet
Paper • 2204.08714 • Published
NAFSSR: Optimized for Qualcomm Devices
NAFNET is designed for lightweight real-time upscaling of images.
This is based on the implementation of NAFSSR 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 |
|---|---|---|---|---|
| PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | QAIRT 2.50, ONNX Runtime 1.30.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | QAIRT 2.50, ONNX Runtime 1.30.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Snapdragon® X2 Elite | QAIRT 2.50, ONNX Runtime 1.30.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Snapdragon® X Elite | QAIRT 2.50, ONNX Runtime 1.30.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Gen 3 Mobile | QAIRT 2.50, ONNX Runtime 1.30.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Gen 1 Mobile | QAIRT 2.50, ONNX Runtime 1.30.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | QAIRT 2.50, ONNX Runtime 1.30.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | QAIRT 2.50, ONNX Runtime 1.30.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | QAIRT 2.50, ONNX Runtime 1.30.0 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | QAIRT 2.50 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® 8 Elite For Galaxy Mobile | QAIRT 2.50 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® X2 Elite | QAIRT 2.50 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® X Elite | QAIRT 2.50 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® 8 Gen 3 Mobile | QAIRT 2.50 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® 8 Gen 1 Mobile | QAIRT 2.50 | Download |
| QNN_CONTEXT_BINARY | float | Qualcomm® Dragonwing™ IQ-8275 | QAIRT 2.50 | Download |
| QNN_CONTEXT_BINARY | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | QAIRT 2.50 | Download |
| QNN_CONTEXT_BINARY | float | Qualcomm® SA8775P | QAIRT 2.50 | Download |
| QNN_CONTEXT_BINARY | float | Qualcomm® Dragonwing™ IQ-9075 | QAIRT 2.50 | Download |
| QNN_CONTEXT_BINARY | float | Qualcomm® SA7255P | QAIRT 2.50 | Download |
| QNN_CONTEXT_BINARY | float | Qualcomm® SA8295P | QAIRT 2.50 | Download |
For more device-specific assets and performance metrics, visit NAFSSR 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 NAFSSR on GitHub for usage instructions.
Model Details
Model Type: Model_use_case.super_resolution
Model Stats:
- Input resolution: 128x128
- Model checkpoint: NAFSSR-S_2x
- Model size (float): 6.2 MB
- Number of parameters: 1.54M
Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |
|---|---|---|---|---|---|---|
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 77.205 ms | 7 - 15 MB | NPU |
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 101.91 ms | 6 - 12 MB | NPU |
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Snapdragon® X2 Elite | 80.612 ms | 7 - 7 MB | NPU |
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Snapdragon® X Elite | 198.15 ms | 32 - 32 MB | NPU |
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Gen 3 Mobile | 130.706 ms | 5 - 13 MB | NPU |
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Gen 1 Mobile | 208.579 ms | 4 - 17 MB | NPU |
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 168.976 ms | 3 - 8 MB | NPU |
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 188.076 ms | 0 - 33 MB | NPU |
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Qualcomm® QCS8450 | 208.579 ms | 4 - 17 MB | NPU |
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 191.628 ms | 5 - 8 MB | NPU |
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 198.15 ms | 32 - 32 MB | NPU |
| NAFSSR | PRECOMPILED_QNN_ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 101.91 ms | 6 - 12 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 77.112 ms | 3 - 11 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Snapdragon® 8 Elite For Galaxy Mobile | 101.772 ms | 1 - 9 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Snapdragon® X2 Elite | 80.989 ms | 0 - 0 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Snapdragon® X Elite | 197.947 ms | 0 - 0 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Snapdragon® 8 Gen 3 Mobile | 130.646 ms | 1 - 8 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Snapdragon® 8 Gen 1 Mobile | 210.202 ms | 1 - 11 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Qualcomm® Dragonwing™ IQ-8275 | 168.546 ms | 0 - 5 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 186.758 ms | 1 - 3 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Qualcomm® SA8650P | 195.284 ms | 1 - 9 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Qualcomm® SA8255P | 195.284 ms | 1 - 9 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Qualcomm® QCS8450 | 210.202 ms | 1 - 11 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Qualcomm® Dragonwing™ IQ-9075 | 190.466 ms | 2 - 6 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Qualcomm® Dragonwing™ IQ-X7181 | 197.947 ms | 0 - 0 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Qualcomm® Dragonwing™ Q-8750 | 101.772 ms | 1 - 9 MB | NPU |
| NAFSSR | QNN_CONTEXT_BINARY | float | Qualcomm® SA8295P | 223.319 ms | 1 - 6 MB | NPU |
License
- The license for the original implementation of NAFSSR 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.