ESRGAN: Optimized for Qualcomm Devices

ESRGAN is a machine learning model that upscales an image with minimal loss in quality.

This is based on the implementation of ESRGAN 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	w8a16	Universal	QAIRT 2.45, ONNX Runtime 1.25.0	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	w8a16	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit ESRGAN 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 ESRGAN on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.super_resolution

Model Stats:

• Model checkpoint: ESRGAN_x4
• Input resolution: 128x128
• Number of parameters: 16.7M
• Model size (float): 63.9 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
ESRGAN	ONNX	float	Snapdragon® X2 Elite	34.373 ms	8 - 8 MB	NPU
ESRGAN	ONNX	float	Snapdragon® X Elite	65.409 ms	38 - 38 MB	NPU
ESRGAN	ONNX	float	Snapdragon® 8 Gen 3 Mobile	49.614 ms	0 - 745 MB	NPU
ESRGAN	ONNX	float	Snapdragon® 8 Gen 1 Mobile	148.068 ms	1 - 735 MB	NPU
ESRGAN	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	67.114 ms	0 - 43 MB	NPU
ESRGAN	ONNX	float	Qualcomm® QCS8450	148.068 ms	1 - 735 MB	NPU
ESRGAN	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	26.907 ms	5 - 357 MB	NPU
ESRGAN	ONNX	float	Snapdragon® 8 Elite Mobile	38.622 ms	8 - 357 MB	NPU
ESRGAN	ONNX	float	Qualcomm® Dragonwing™ Q-8750	38.622 ms	8 - 357 MB	NPU
ESRGAN	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	65.409 ms	38 - 38 MB	NPU
ESRGAN	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	107.039 ms	7 - 52 MB	NPU
ESRGAN	ONNX	w8a16	Snapdragon® X2 Elite	22.012 ms	4 - 4 MB	NPU
ESRGAN	ONNX	w8a16	Snapdragon® X Elite	43.476 ms	26 - 26 MB	NPU
ESRGAN	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	32.056 ms	3 - 1130 MB	NPU
ESRGAN	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	84.013 ms	4 - 1214 MB	NPU
ESRGAN	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	323.426 ms	3 - 6 MB	NPU
ESRGAN	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	42.546 ms	0 - 32 MB	NPU
ESRGAN	ONNX	w8a16	Qualcomm® QCS8450	84.013 ms	4 - 1214 MB	NPU
ESRGAN	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	89.842 ms	3 - 781 MB	NPU
ESRGAN	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	16.936 ms	3 - 1090 MB	NPU
ESRGAN	ONNX	w8a16	Qualcomm® Dragonwing™ Q-6690	1077.961 ms	0 - 725 MB	NPU
ESRGAN	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	45.393 ms	3 - 48 MB	NPU
ESRGAN	ONNX	w8a16	Snapdragon® 8 Elite Mobile	26.47 ms	3 - 923 MB	NPU
ESRGAN	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	89.842 ms	3 - 781 MB	NPU
ESRGAN	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	26.47 ms	3 - 923 MB	NPU
ESRGAN	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	43.476 ms	26 - 26 MB	NPU
ESRGAN	QNN_DLC	float	Snapdragon® X2 Elite	34.365 ms	0 - 0 MB	NPU
ESRGAN	QNN_DLC	float	Snapdragon® X Elite	64.979 ms	0 - 0 MB	NPU
ESRGAN	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	49.483 ms	0 - 706 MB	NPU
ESRGAN	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	130.701 ms	0 - 708 MB	NPU
ESRGAN	QNN_DLC	float	Qualcomm® QCS8275	452.056 ms	1 - 353 MB	NPU
ESRGAN	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	65.392 ms	0 - 2 MB	NPU
ESRGAN	QNN_DLC	float	Qualcomm® SA8775P	105.496 ms	1 - 352 MB	NPU
ESRGAN	QNN_DLC	float	Qualcomm® SA8650P	105.496 ms	1 - 352 MB	NPU
ESRGAN	QNN_DLC	float	Qualcomm® SA8255P	105.496 ms	1 - 352 MB	NPU
ESRGAN	QNN_DLC	float	Qualcomm® QCS8450	130.701 ms	0 - 708 MB	NPU
ESRGAN	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	26.21 ms	0 - 335 MB	NPU
ESRGAN	QNN_DLC	float	Qualcomm® SA7255P	452.056 ms	1 - 353 MB	NPU
ESRGAN	QNN_DLC	float	Snapdragon® 8 Elite Mobile	37.79 ms	0 - 335 MB	NPU
ESRGAN	QNN_DLC	float	Qualcomm® SA8295P	111.355 ms	0 - 359 MB	NPU
ESRGAN	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	37.79 ms	0 - 335 MB	NPU
ESRGAN	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	64.979 ms	0 - 0 MB	NPU
ESRGAN	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	106.612 ms	0 - 5 MB	NPU
ESRGAN	QNN_DLC	w8a16	Snapdragon® X2 Elite	21.863 ms	0 - 0 MB	NPU
ESRGAN	QNN_DLC	w8a16	Snapdragon® X Elite	43.234 ms	0 - 0 MB	NPU
ESRGAN	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	31.539 ms	0 - 1037 MB	NPU
ESRGAN	QNN_DLC	w8a16	Snapdragon® 8 Gen 1 Mobile	81.224 ms	1 - 1115 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS6490	320.157 ms	2 - 5 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® QCS8275	132.843 ms	1 - 672 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	42.106 ms	0 - 610 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® SA8775P	37.667 ms	0 - 672 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® SA8650P	37.667 ms	0 - 672 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® SA8255P	37.667 ms	0 - 672 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® QCS8450	81.224 ms	1 - 1115 MB	NPU
ESRGAN	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	88.736 ms	0 - 697 MB	NPU
ESRGAN	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	16.471 ms	0 - 1003 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	1156.873 ms	0 - 635 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	38.041 ms	2 - 5 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® SA7255P	132.843 ms	1 - 672 MB	NPU
ESRGAN	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	28.069 ms	0 - 843 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® SA8295P	65.063 ms	0 - 717 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	88.736 ms	0 - 697 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	28.069 ms	0 - 843 MB	NPU
ESRGAN	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	43.234 ms	0 - 0 MB	NPU
ESRGAN	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	51.37 ms	3 - 747 MB	NPU
ESRGAN	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	128.931 ms	3 - 746 MB	NPU
ESRGAN	TFLITE	float	Qualcomm® QCS8275	451.996 ms	4 - 396 MB	NPU
ESRGAN	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	65.104 ms	3 - 433 MB	NPU
ESRGAN	TFLITE	float	Qualcomm® SA8775P	105.572 ms	3 - 396 MB	NPU
ESRGAN	TFLITE	float	Qualcomm® SA8650P	105.572 ms	3 - 396 MB	NPU
ESRGAN	TFLITE	float	Qualcomm® SA8255P	105.572 ms	3 - 396 MB	NPU
ESRGAN	TFLITE	float	Qualcomm® QCS8450	128.931 ms	3 - 746 MB	NPU
ESRGAN	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	26.204 ms	1 - 364 MB	NPU
ESRGAN	TFLITE	float	Qualcomm® SA7255P	451.996 ms	4 - 396 MB	NPU
ESRGAN	TFLITE	float	Snapdragon® 8 Elite Mobile	38.455 ms	14 - 377 MB	NPU
ESRGAN	TFLITE	float	Qualcomm® SA8295P	111.369 ms	3 - 392 MB	NPU
ESRGAN	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	38.455 ms	14 - 377 MB	NPU
ESRGAN	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	108.259 ms	3 - 46 MB	NPU

License

• The license for the original implementation of ESRGAN can be found here.

References

• ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks
• Source Model Implementation

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.