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. 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	w8a8	Universal	QAIRT 2.45, ONNX Runtime 1.25.0	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	w8a8	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download
TFLITE	w8a8	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit DnCNN 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 DnCNN on GitHub 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® X2 Elite	4.046 ms	1 - 1 MB	NPU
DnCNN	ONNX	float	Snapdragon® X Elite	7.147 ms	0 - 0 MB	NPU
DnCNN	ONNX	float	Snapdragon® 8 Gen 3 Mobile	5.169 ms	1 - 178 MB	NPU
DnCNN	ONNX	float	Snapdragon® 8 Gen 1 Mobile	13.769 ms	1 - 179 MB	NPU
DnCNN	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.88 ms	1 - 3 MB	NPU
DnCNN	ONNX	float	Qualcomm® QCS8450	13.769 ms	1 - 179 MB	NPU
DnCNN	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	14.207 ms	1 - 4 MB	NPU
DnCNN	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	3.205 ms	0 - 144 MB	NPU
DnCNN	ONNX	float	Snapdragon® 8 Elite Mobile	4.122 ms	0 - 139 MB	NPU
DnCNN	ONNX	float	Qualcomm® Dragonwing™ Q-8750	4.122 ms	0 - 139 MB	NPU
DnCNN	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	7.147 ms	0 - 0 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® X2 Elite	1.047 ms	1 - 1 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® X Elite	1.87 ms	0 - 0 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	1.338 ms	0 - 49 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® 8 Gen 1 Mobile	2.386 ms	0 - 52 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ QCS6490	9.407 ms	0 - 3 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ QCS8550 (Proxy)	1.79 ms	0 - 78 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® QCS8450	2.386 ms	0 - 52 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® 8 Elite Mobile	1.224 ms	0 - 34 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	3.259 ms	0 - 143 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.811 ms	0 - 31 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ IQ-9075	1.888 ms	0 - 3 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ Q-6690	40.072 ms	0 - 143 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ Q-7790	3.259 ms	0 - 143 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ Q-8750	1.224 ms	0 - 34 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ IQ-X7181	1.87 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® X2 Elite	4.165 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® X Elite	7.282 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	5.074 ms	0 - 177 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	13.432 ms	0 - 177 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® QCS8275	55.975 ms	0 - 141 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.698 ms	0 - 6 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® SA8775P	13.864 ms	0 - 143 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® SA8650P	13.864 ms	0 - 143 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® SA8255P	13.864 ms	0 - 143 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® QCS8450	13.432 ms	0 - 177 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	13.738 ms	2 - 4 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	2.967 ms	0 - 148 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® SA7255P	55.975 ms	0 - 141 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® 8 Elite Mobile	4.029 ms	0 - 141 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® SA8295P	15.319 ms	0 - 139 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	4.029 ms	0 - 141 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	7.282 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® X2 Elite	1.182 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® X Elite	2.005 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	1.335 ms	0 - 45 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® 8 Gen 1 Mobile	2.379 ms	0 - 50 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ QCS6490	9.235 ms	0 - 2 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® QCS8275	7.59 ms	0 - 28 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ QCS8550 (Proxy)	1.788 ms	0 - 1 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® SA8775P	2.007 ms	0 - 30 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® SA8650P	2.007 ms	0 - 30 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® SA8255P	2.007 ms	0 - 30 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® QCS8450	2.379 ms	0 - 50 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® 8 Elite Mobile	1.212 ms	0 - 29 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	3.251 ms	0 - 140 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.81 ms	0 - 28 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ IQ-9075	1.868 ms	2 - 4 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ Q-6690	39.54 ms	0 - 140 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® SA7255P	7.59 ms	0 - 28 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® SA8295P	4.232 ms	0 - 26 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ Q-7790	3.251 ms	0 - 140 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ Q-8750	1.212 ms	0 - 29 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ IQ-X7181	2.005 ms	0 - 0 MB	NPU
DnCNN	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	5.204 ms	0 - 179 MB	NPU
DnCNN	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	13.885 ms	0 - 179 MB	NPU
DnCNN	TFLITE	float	Qualcomm® QCS8275	56.399 ms	0 - 142 MB	NPU
DnCNN	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.905 ms	0 - 2 MB	NPU
DnCNN	TFLITE	float	Qualcomm® SA8775P	14.16 ms	0 - 145 MB	NPU
DnCNN	TFLITE	float	Qualcomm® SA8650P	14.16 ms	0 - 145 MB	NPU
DnCNN	TFLITE	float	Qualcomm® SA8255P	14.16 ms	0 - 145 MB	NPU
DnCNN	TFLITE	float	Qualcomm® QCS8450	13.885 ms	0 - 179 MB	NPU
DnCNN	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	14.3 ms	0 - 4 MB	NPU
DnCNN	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	3.163 ms	0 - 149 MB	NPU
DnCNN	TFLITE	float	Qualcomm® SA7255P	56.399 ms	0 - 142 MB	NPU
DnCNN	TFLITE	float	Snapdragon® 8 Elite Mobile	4.128 ms	0 - 143 MB	NPU
DnCNN	TFLITE	float	Qualcomm® SA8295P	15.582 ms	0 - 141 MB	NPU
DnCNN	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	4.128 ms	0 - 143 MB	NPU
DnCNN	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	1.295 ms	0 - 47 MB	NPU
DnCNN	TFLITE	w8a8	Snapdragon® 8 Gen 1 Mobile	2.33 ms	0 - 52 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ QCS6490	9.344 ms	0 - 3 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® QCS8275	7.504 ms	0 - 29 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ QCS8550 (Proxy)	1.707 ms	0 - 76 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® SA8775P	1.984 ms	0 - 31 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® SA8650P	1.984 ms	0 - 31 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® SA8255P	1.984 ms	0 - 31 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® QCS8450	2.33 ms	0 - 52 MB	NPU
DnCNN	TFLITE	w8a8	Snapdragon® 8 Elite Mobile	1.183 ms	0 - 33 MB	NPU
DnCNN	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	3.214 ms	0 - 141 MB	NPU
DnCNN	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.77 ms	0 - 29 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ IQ-9075	1.817 ms	0 - 3 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ Q-6690	38.938 ms	0 - 140 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® SA7255P	7.504 ms	0 - 29 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® SA8295P	4.19 ms	0 - 27 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ Q-7790	3.214 ms	0 - 141 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ Q-8750	1.183 ms	0 - 33 MB	NPU

License

The license for the original implementation of DnCNN 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.

Downloads last month: -; Downloads are not tracked for this model. How to track

Paper for qualcomm/DnCNN

Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising

Paper • 1608.03981 • Published Aug 13, 2016