---
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/main/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 2.42, ONNX Runtime 1.25.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/dncnn/releases/v0.55.0/dncnn-onnx-float.zip)
| ONNX | w8a8 | Universal | QAIRT 2.42, ONNX Runtime 1.25.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/dncnn/releases/v0.55.0/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/v0.55.0/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/v0.55.0/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/v0.55.0/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/v0.55.0/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/main/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/main/src/qai_hub_models/models/dncnn) for usage instructions.

## Model Details

**Model Type:** Model_use_case.image_editing

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

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| DnCNN | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.118 ms | 0 - 143 MB | NPU
| DnCNN | ONNX | float | Snapdragon® X2 Elite | 4.042 ms | 212 - 212 MB | NPU
| DnCNN | ONNX | float | Snapdragon® X Elite | 7.137 ms | 180 - 180 MB | NPU
| DnCNN | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 5.163 ms | 1 - 180 MB | NPU
| DnCNN | ONNX | float | Qualcomm® QCS8550 (Proxy) | 6.882 ms | 1 - 15 MB | NPU
| DnCNN | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 4.13 ms | 0 - 143 MB | NPU
| DnCNN | ONNX | float | Qualcomm® QCS9075 | 14.432 ms | 0 - 46 MB | NPU
| DnCNN | ONNX | float | Qualcomm® QCS8750 | 4.13 ms | 0 - 143 MB | NPU
| DnCNN | ONNX | float | Qualcomm® QCS7181 | 7.137 ms | 180 - 180 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.753 ms | 0 - 34 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® X2 Elite | 1.044 ms | 212 - 212 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® X Elite | 1.862 ms | 159 - 159 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.33 ms | 0 - 49 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® QCS6490 | 131.428 ms | 60 - 63 MB | CPU
| DnCNN | ONNX | w8a8 | Qualcomm® QCS8550 (Proxy) | 1.787 ms | 0 - 46 MB | NPU
| DnCNN | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 160.653 ms | 62 - 69 MB | CPU
| DnCNN | ONNX | w8a8 | Qualcomm® QCM6690 | 192.027 ms | 59 - 67 MB | CPU
| DnCNN | ONNX | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 1.218 ms | 0 - 30 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® QCS9075 | 1.942 ms | 0 - 46 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® QCS7790 | 160.653 ms | 62 - 69 MB | CPU
| DnCNN | ONNX | w8a8 | Qualcomm® QCS8750 | 1.218 ms | 0 - 30 MB | NPU
| DnCNN | ONNX | w8a8 | Qualcomm® QCS7181 | 1.862 ms | 159 - 159 MB | NPU
| DnCNN | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 2.968 ms | 0 - 146 MB | NPU
| DnCNN | QNN_DLC | float | Snapdragon® X2 Elite | 4.17 ms | 0 - 0 MB | NPU
| DnCNN | QNN_DLC | float | Snapdragon® X Elite | 7.194 ms | 0 - 0 MB | NPU
| DnCNN | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 5.046 ms | 0 - 176 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® QCS8275 | 55.973 ms | 0 - 141 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 6.712 ms | 0 - 2 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® SA8775P | 13.874 ms | 0 - 143 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® SA8650P | 13.874 ms | 0 - 143 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® SA8255P | 13.874 ms | 0 - 143 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 13.453 ms | 0 - 178 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® SA8295P | 15.31 ms | 0 - 139 MB | NPU
| DnCNN | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 4.03 ms | 0 - 145 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® SA7255P | 55.973 ms | 0 - 141 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® QCS9075 | 13.877 ms | 0 - 2 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® QCS8750 | 4.03 ms | 0 - 145 MB | NPU
| DnCNN | QNN_DLC | float | Qualcomm® QCS7181 | 7.194 ms | 0 - 0 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.754 ms | 0 - 31 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 1.207 ms | 0 - 0 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® X Elite | 1.998 ms | 0 - 0 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.336 ms | 0 - 46 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® QCS6490 | 7.555 ms | 2 - 4 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® QCS8275 | 7.576 ms | 0 - 29 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® QCS8550 (Proxy) | 1.781 ms | 0 - 2 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® SA8775P | 2.006 ms | 0 - 30 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® SA8650P | 2.006 ms | 0 - 30 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® SA8255P | 2.006 ms | 0 - 30 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® SA7255P | 7.576 ms | 0 - 29 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 3.257 ms | 0 - 140 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® SA8295P | 4.226 ms | 0 - 26 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® QCM6690 | 39.089 ms | 0 - 140 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 1.213 ms | 0 - 29 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® QCS9075 | 1.926 ms | 0 - 2 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® QCS8450 (Proxy) | 2.352 ms | 0 - 47 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® QCS7790 | 3.257 ms | 0 - 140 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® QCS8750 | 1.213 ms | 0 - 29 MB | NPU
| DnCNN | QNN_DLC | w8a8 | Qualcomm® QCS7181 | 1.998 ms | 0 - 0 MB | NPU
| DnCNN | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.088 ms | 0 - 148 MB | NPU
| DnCNN | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 5.186 ms | 0 - 178 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® QCS8275 | 56.393 ms | 0 - 142 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 6.925 ms | 0 - 2 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® SA8775P | 14.164 ms | 0 - 145 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® SA8650P | 14.164 ms | 0 - 145 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® SA8255P | 14.164 ms | 0 - 145 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 13.91 ms | 1 - 175 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® SA8295P | 15.608 ms | 0 - 141 MB | NPU
| DnCNN | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 4.124 ms | 0 - 149 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® SA7255P | 56.393 ms | 0 - 142 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® QCS9075 | 14.438 ms | 0 - 4 MB | NPU
| DnCNN | TFLITE | float | Qualcomm® QCS8750 | 4.124 ms | 0 - 149 MB | NPU
| DnCNN | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.718 ms | 0 - 31 MB | NPU
| DnCNN | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.3 ms | 0 - 46 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® QCS6490 | 7.819 ms | 0 - 3 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® QCS8275 | 7.519 ms | 0 - 28 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® QCS8550 (Proxy) | 1.724 ms | 0 - 4 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® SA8775P | 1.984 ms | 0 - 30 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® SA8650P | 1.984 ms | 0 - 30 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® SA8255P | 1.984 ms | 0 - 30 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® SA7255P | 7.519 ms | 0 - 28 MB | NPU
| DnCNN | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 3.214 ms | 0 - 141 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® SA8295P | 4.175 ms | 0 - 26 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® QCM6690 | 39.027 ms | 0 - 141 MB | NPU
| DnCNN | TFLITE | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 1.175 ms | 0 - 29 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® QCS9075 | 1.846 ms | 0 - 3 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® QCS8450 (Proxy) | 2.355 ms | 0 - 48 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® QCS7790 | 3.214 ms | 0 - 141 MB | NPU
| DnCNN | TFLITE | w8a8 | Qualcomm® QCS8750 | 1.175 ms | 0 - 29 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).