README.md

---
library_name: pytorch
license: other
tags:
- real_time
- android
pipeline_tag: image-to-image

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/dncnn/web-assets/model_demo.png)

# DnCNN: Optimized for Qualcomm Devices

DnCNN is a 17-layer denoising convolutional neural network that uses residual learning to remove Gaussian noise (sigma=25) from grayscale images. The network predicts the noise residual and subtracts it from the input to produce a clean image.

This is based on the implementation of DnCNN found [here](https://github.com/cszn/KAIR).
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.60.0/src/qai_hub_models/models/dncnn) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).

Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) 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](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/dncnn/releases/v0.60.0/dncnn-onnx-float.zip)
| ONNX | w8a8 | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/dncnn/releases/v0.60.0/dncnn-onnx-w8a8.zip)
| QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/dncnn/releases/v0.60.0/dncnn-qnn_dlc-float.zip)
| QNN_DLC | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/dncnn/releases/v0.60.0/dncnn-qnn_dlc-w8a8.zip)
| TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/dncnn/releases/v0.60.0/dncnn-tflite-float.zip)
| TFLITE | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/dncnn/releases/v0.60.0/dncnn-tflite-w8a8.zip)

For more device-specific assets and performance metrics, visit **[DnCNN on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/dncnn)**.


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.60.0/src/qai_hub_models/models/dncnn) 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 [DnCNN on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.60.0/src/qai_hub_models/models/dncnn) for usage instructions.

## Model Details

**Model Type:** Model_use_case.image_editing

**Model Stats:**
- Input resolution: 256x256
- Model checkpoint: dncnn_25
- Model size (float): 2.12 MB
- Model size (w8a8): 581 KB
- Number of parameters: 555K

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| DnCNN | ONNX | float | Snapdragon® X2 Elite | 4.043 ms | 1 - 1 MB | NPU
| DnCNN | ONNX | float | Snapdragon® X Elite | 7.136 ms | 0 - 0 MB | NPU
| DnCNN | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 5.185 ms | 19 - 195 MB | NPU
| DnCNN | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 13.757 ms | 1 - 179 MB | NPU
| DnCNN | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 16.317 ms | 1 - 4 MB | NPU
| DnCNN | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 6.874 ms | 1 - 3 MB | NPU
| DnCNN | ONNX | float | Qualcomm® QCS8450 | 13.757 ms | 1 - 179 MB | NPU
| DnCNN | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 14.202 ms | 1 - 4 MB | NPU
| DnCNN | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 7.136 ms | 0 - 0 MB | NPU
| DnCNN | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 4.126 ms | 0 - 142 MB | NPU
| DnCNN | ONNX | float | Snapdragon® 8 Elite Mobile | 4.126 ms | 0 - 142 MB | NPU
| DnCNN | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.197 ms | 0 - 146 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® X2 Elite | 1.043 ms | 1 - 1 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® X Elite | 1.873 ms | 0 - 0 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.336 ms | 0 - 47 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 2.371 ms | 0 - 52 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 9.5 ms | 0 - 3 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 2.205 ms | 0 - 4 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.788 ms | 0 - 2 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® QCS8450 | 2.371 ms | 0 - 52 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 1.898 ms | 0 - 3 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 1.873 ms | 0 - 0 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 39.004 ms | 2 - 145 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 3.262 ms | 0 - 144 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 1.209 ms | 0 - 31 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® 8 Elite Mobile | 1.209 ms | 0 - 31 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.812 ms | 0 - 32 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 3.262 ms | 0 - 144 MB | NPU
| DnCNN | QNN_DLC | float | Snapdragon® X2 Elite | 4.166 ms | 0 - 0 MB | NPU
| DnCNN | QNN_DLC | float | Snapdragon® X Elite | 7.217 ms | 0 - 0 MB | NPU
| DnCNN | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 5.085 ms | 0 - 173 MB | NPU
| DnCNN | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 13.451 ms | 0 - 177 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 15.809 ms | 0 - 3 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 55.992 ms | 0 - 141 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 6.729 ms | 0 - 2 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® SA8775P | 13.869 ms | 0 - 144 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® SA8650P | 13.869 ms | 0 - 144 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® SA8255P | 13.869 ms | 0 - 144 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® QCS8450 | 13.451 ms | 0 - 177 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 13.982 ms | 0 - 2 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 7.217 ms | 0 - 0 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 4.037 ms | 0 - 141 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® SA7255P | 55.992 ms | 0 - 141 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® SA8295P | 15.313 ms | 0 - 139 MB | NPU
| DnCNN | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 4.037 ms | 0 - 141 MB | NPU
| DnCNN | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.05 ms | 0 - 149 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 1.176 ms | 0 - 0 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® X Elite | 2.007 ms | 0 - 0 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.323 ms | 0 - 46 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 2.514 ms | 0 - 51 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 9.419 ms | 0 - 2 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 2.183 ms | 0 - 2 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 7.581 ms | 0 - 28 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.777 ms | 0 - 3 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® SA8775P | 1.999 ms | 0 - 30 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® SA8650P | 1.999 ms | 0 - 30 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® SA8255P | 1.999 ms | 0 - 30 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 2.514 ms | 0 - 51 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 2.02 ms | 2 - 4 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 2.007 ms | 0 - 0 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 38.849 ms | 0 - 140 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 3.256 ms | 0 - 140 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 1.221 ms | 0 - 28 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® SA7255P | 7.581 ms | 0 - 28 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® SA8295P | 4.226 ms | 0 - 26 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® 8 Elite Mobile | 1.221 ms | 0 - 28 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.805 ms | 0 - 28 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 3.256 ms | 0 - 140 MB | NPU
| DnCNN | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 5.2 ms | 0 - 179 MB | NPU
| DnCNN | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 14.005 ms | 0 - 180 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 16.242 ms | 0 - 5 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 56.39 ms | 0 - 142 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 6.928 ms | 0 - 2 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® SA8775P | 14.161 ms | 0 - 144 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® SA8650P | 14.161 ms | 0 - 144 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® SA8255P | 14.161 ms | 0 - 144 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® QCS8450 | 14.005 ms | 0 - 180 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 14.318 ms | 0 - 4 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 4.163 ms | 0 - 147 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® SA7255P | 56.39 ms | 0 - 142 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® SA8295P | 15.612 ms | 0 - 141 MB | NPU
| DnCNN | TFLITE | float | Snapdragon® 8 Elite Mobile | 4.163 ms | 0 - 147 MB | NPU
| DnCNN | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.173 ms | 0 - 146 MB | NPU
| DnCNN | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.302 ms | 0 - 47 MB | NPU
| DnCNN | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 2.343 ms | 0 - 52 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 9.451 ms | 0 - 3 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 2.14 ms | 0 - 3 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 7.522 ms | 0 - 29 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.711 ms | 0 - 72 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® SA8775P | 1.98 ms | 0 - 30 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® SA8650P | 1.98 ms | 0 - 30 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® SA8255P | 1.98 ms | 0 - 30 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® QCS8450 | 2.343 ms | 0 - 52 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 1.819 ms | 0 - 3 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 38.909 ms | 0 - 141 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 3.215 ms | 0 - 141 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 1.19 ms | 0 - 30 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® SA7255P | 7.522 ms | 0 - 29 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® SA8295P | 4.183 ms | 0 - 26 MB | NPU
| DnCNN | TFLITE | w8a8 | Snapdragon® 8 Elite Mobile | 1.19 ms | 0 - 30 MB | NPU
| DnCNN | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.771 ms | 0 - 30 MB | NPU
| DnCNN | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 3.215 ms | 0 - 141 MB | NPU

## License
* The license for the original implementation of DnCNN can be found
  [here](https://github.com/cszn/KAIR/blob/master/LICENSE).

## References
* [Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising](https://arxiv.org/abs/1608.03981)
* [Source Model Implementation](https://github.com/cszn/KAIR)

## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).