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 2.50, ONNX Runtime 1.27.1	Download
ONNX	w8a8	Universal	QAIRT 2.50, ONNX Runtime 1.27.1	Download
QNN_DLC	float	Universal	QAIRT 2.50	Download
QNN_DLC	w8a8	Universal	QAIRT 2.50	Download
TFLITE	float	Universal	QAIRT 2.50	Download
TFLITE	w8a8	Universal	QAIRT 2.50	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:

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® 8 Elite Gen 5 For Galaxy Mobile	3.176 ms	0 - 137 MB	NPU
DnCNN	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	4.146 ms	0 - 135 MB	NPU
DnCNN	ONNX	float	Snapdragon® X2 Elite	4.004 ms	1 - 1 MB	NPU
DnCNN	ONNX	float	Snapdragon® X Elite	7.172 ms	0 - 0 MB	NPU
DnCNN	ONNX	float	Snapdragon® 8 Gen 3 Mobile	5.157 ms	1 - 170 MB	NPU
DnCNN	ONNX	float	Snapdragon® 8 Gen 1 Mobile	13.062 ms	1 - 176 MB	NPU
DnCNN	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	16.066 ms	1 - 5 MB	NPU
DnCNN	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.919 ms	1 - 3 MB	NPU
DnCNN	ONNX	float	Qualcomm® QCS8450	13.062 ms	1 - 176 MB	NPU
DnCNN	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	14.355 ms	1 - 4 MB	NPU
DnCNN	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	7.172 ms	0 - 0 MB	NPU
DnCNN	ONNX	float	Qualcomm® Dragonwing™ Q-8750	4.146 ms	0 - 135 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	0.81 ms	0 - 35 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	1.212 ms	0 - 34 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® X2 Elite	1.036 ms	1 - 1 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® X Elite	1.872 ms	0 - 0 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	1.327 ms	0 - 52 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® 8 Gen 1 Mobile	2.39 ms	0 - 58 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ QCS6490	7.84 ms	0 - 3 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ IQ-8275	2.196 ms	0 - 4 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ QCS8550 (Proxy)	1.793 ms	0 - 2 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® QCS8450	2.39 ms	0 - 58 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ IQ-9075	1.893 ms	0 - 3 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ IQ-X7181	1.872 ms	0 - 0 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ Q-6690	38.886 ms	0 - 141 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ Q-7790	3.281 ms	0 - 139 MB	NPU
DnCNN	ONNX	w8a8	Qualcomm® Dragonwing™ Q-8750	1.212 ms	0 - 34 MB	NPU
DnCNN	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	3.281 ms	0 - 139 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	3.039 ms	0 - 143 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	4.031 ms	0 - 137 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® X2 Elite	4.192 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® X Elite	7.33 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	5.072 ms	0 - 171 MB	NPU
DnCNN	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	13.088 ms	0 - 175 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	16.013 ms	0 - 4 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.76 ms	0 - 2 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® SA8775P	13.901 ms	0 - 141 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® SA8650P	13.901 ms	0 - 141 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® SA8255P	13.901 ms	0 - 141 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® QCS8450	13.088 ms	0 - 175 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	14.31 ms	0 - 3 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	7.33 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	4.031 ms	0 - 137 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® SA7255P	55.996 ms	0 - 137 MB	NPU
DnCNN	QNN_DLC	float	Qualcomm® SA8295P	15.172 ms	0 - 136 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	0.805 ms	0 - 33 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	1.228 ms	0 - 36 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® X2 Elite	1.205 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® X Elite	1.991 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	1.337 ms	0 - 50 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® 8 Gen 1 Mobile	2.355 ms	0 - 55 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ QCS6490	7.514 ms	2 - 4 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ IQ-8275	2.181 ms	0 - 3 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ QCS8550 (Proxy)	1.798 ms	0 - 1 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® SA8775P	1.999 ms	0 - 34 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® SA8650P	1.999 ms	0 - 34 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® SA8255P	1.999 ms	0 - 34 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® QCS8450	2.355 ms	0 - 55 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ IQ-9075	1.87 ms	0 - 2 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ IQ-X7181	1.991 ms	0 - 0 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ Q-6690	39.527 ms	2 - 139 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ Q-7790	3.271 ms	0 - 135 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® Dragonwing™ Q-8750	1.228 ms	0 - 36 MB	NPU
DnCNN	QNN_DLC	w8a8	Qualcomm® SA8295P	4.084 ms	0 - 29 MB	NPU
DnCNN	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	3.271 ms	0 - 135 MB	NPU
DnCNN	TFLITE	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	3.069 ms	0 - 141 MB	NPU
DnCNN	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	4.035 ms	0 - 142 MB	NPU
DnCNN	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	5.092 ms	0 - 170 MB	NPU
DnCNN	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	12.976 ms	0 - 173 MB	NPU
DnCNN	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	15.819 ms	0 - 5 MB	NPU
DnCNN	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.712 ms	0 - 2 MB	NPU
DnCNN	TFLITE	float	Qualcomm® SA8775P	13.85 ms	0 - 141 MB	NPU
DnCNN	TFLITE	float	Qualcomm® SA8650P	13.85 ms	0 - 141 MB	NPU
DnCNN	TFLITE	float	Qualcomm® SA8255P	13.85 ms	0 - 141 MB	NPU
DnCNN	TFLITE	float	Qualcomm® QCS8450	12.976 ms	0 - 173 MB	NPU
DnCNN	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	13.67 ms	0 - 4 MB	NPU
DnCNN	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	4.035 ms	0 - 142 MB	NPU
DnCNN	TFLITE	float	Qualcomm® SA7255P	55.989 ms	0 - 138 MB	NPU
DnCNN	TFLITE	float	Qualcomm® SA8295P	15.339 ms	0 - 137 MB	NPU
DnCNN	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	0.773 ms	0 - 32 MB	NPU
DnCNN	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	1.185 ms	0 - 36 MB	NPU
DnCNN	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	1.296 ms	0 - 50 MB	NPU
DnCNN	TFLITE	w8a8	Snapdragon® 8 Gen 1 Mobile	2.328 ms	0 - 56 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ QCS6490	7.549 ms	0 - 3 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ IQ-8275	2.13 ms	0 - 4 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ QCS8550 (Proxy)	1.71 ms	0 - 17 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® QCS8450	2.328 ms	0 - 56 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ IQ-9075	1.816 ms	0 - 3 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ Q-6690	38.741 ms	0 - 138 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ Q-7790	3.218 ms	0 - 137 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® Dragonwing™ Q-8750	1.185 ms	0 - 36 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® SA7255P	7.502 ms	0 - 33 MB	NPU
DnCNN	TFLITE	w8a8	Qualcomm® SA8295P	4.071 ms	0 - 30 MB	NPU
DnCNN	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	3.218 ms	0 - 137 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