WideResNet50: Optimized for Qualcomm Devices

WideResNet50 is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.

This is based on the implementation of WideResNet50 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 2.42, ONNX Runtime 1.24.3	Download
ONNX	w8a8	Universal	QAIRT 2.42, ONNX Runtime 1.24.3	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 WideResNet50 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 WideResNet50 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_classification

Model Stats:

• Model checkpoint: Imagenet
• Input resolution: 224x224
• Number of parameters: 68.9M
• Model size (float): 263 MB
• Model size (w8a8): 66.6 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
WideResNet50	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	2.635 ms	1 - 189 MB	NPU
WideResNet50	ONNX	float	Snapdragon® 8 Elite Mobile	2.813 ms	0 - 178 MB	NPU
WideResNet50	ONNX	float	Snapdragon® X2 Elite	2.2 ms	132 - 132 MB	NPU
WideResNet50	ONNX	float	Snapdragon® X Elite	4.44 ms	132 - 132 MB	NPU
WideResNet50	ONNX	float	Snapdragon® X Elite	4.44 ms	132 - 132 MB	NPU
WideResNet50	ONNX	float	Snapdragon® 8 Gen 3 Mobile	3.463 ms	0 - 225 MB	NPU
WideResNet50	ONNX	float	Qualcomm® QCS8550 (Proxy)	4.558 ms	1 - 3 MB	NPU
WideResNet50	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	2.813 ms	0 - 178 MB	NPU
WideResNet50	ONNX	float	Qualcomm® QCS9075	6.771 ms	0 - 4 MB	NPU
WideResNet50	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	1.153 ms	0 - 52 MB	NPU
WideResNet50	ONNX	w8a8	Snapdragon® 8 Elite Mobile	1.228 ms	0 - 53 MB	NPU
WideResNet50	ONNX	w8a8	Snapdragon® X2 Elite	0.812 ms	66 - 66 MB	NPU
WideResNet50	ONNX	w8a8	Snapdragon® X Elite	1.799 ms	66 - 66 MB	NPU
WideResNet50	ONNX	w8a8	Snapdragon® X Elite	1.799 ms	66 - 66 MB	NPU
WideResNet50	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	1.409 ms	0 - 131 MB	NPU
WideResNet50	ONNX	w8a8	Qualcomm® QCS6490	81.996 ms	10 - 107 MB	CPU
WideResNet50	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	1.855 ms	0 - 83 MB	NPU
WideResNet50	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	57.376 ms	9 - 18 MB	CPU
WideResNet50	ONNX	w8a8	Qualcomm® QCM6690	61.83 ms	6 - 14 MB	CPU
WideResNet50	ONNX	w8a8	Qualcomm® QCS9075	1.928 ms	0 - 3 MB	NPU
WideResNet50	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	1.228 ms	0 - 53 MB	NPU
WideResNet50	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	57.376 ms	9 - 18 MB	CPU
WideResNet50	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	2.56 ms	1 - 171 MB	NPU
WideResNet50	QNN_DLC	float	Snapdragon® 8 Elite Mobile	2.906 ms	0 - 175 MB	NPU
WideResNet50	QNN_DLC	float	Snapdragon® X2 Elite	2.572 ms	1 - 1 MB	NPU
WideResNet50	QNN_DLC	float	Snapdragon® X Elite	4.686 ms	1 - 1 MB	NPU
WideResNet50	QNN_DLC	float	Snapdragon® X Elite	4.686 ms	1 - 1 MB	NPU
WideResNet50	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	3.634 ms	0 - 227 MB	NPU
WideResNet50	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	4.835 ms	0 - 15 MB	NPU
WideResNet50	QNN_DLC	float	Qualcomm® SA8775P	7.162 ms	1 - 169 MB	NPU
WideResNet50	QNN_DLC	float	Qualcomm® SA8775P	7.162 ms	1 - 169 MB	NPU
WideResNet50	QNN_DLC	float	Qualcomm® SA8775P	7.162 ms	1 - 169 MB	NPU
WideResNet50	QNN_DLC	float	Qualcomm® SA7255P	24.51 ms	1 - 174 MB	NPU
WideResNet50	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	9.204 ms	0 - 199 MB	NPU
WideResNet50	QNN_DLC	float	Qualcomm® SA8295P	7.827 ms	1 - 152 MB	NPU
WideResNet50	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	2.906 ms	0 - 175 MB	NPU
WideResNet50	QNN_DLC	float	Qualcomm® QCS9075	7.064 ms	1 - 3 MB	NPU
WideResNet50	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	1.083 ms	0 - 56 MB	NPU
WideResNet50	QNN_DLC	w8a8	Snapdragon® 8 Elite Mobile	1.162 ms	0 - 53 MB	NPU
WideResNet50	QNN_DLC	w8a8	Snapdragon® X2 Elite	0.883 ms	0 - 0 MB	NPU
WideResNet50	QNN_DLC	w8a8	Snapdragon® X Elite	1.793 ms	0 - 0 MB	NPU
WideResNet50	QNN_DLC	w8a8	Snapdragon® X Elite	1.793 ms	0 - 0 MB	NPU
WideResNet50	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	1.415 ms	0 - 126 MB	NPU
WideResNet50	QNN_DLC	w8a8	Qualcomm® QCS6490	7.707 ms	0 - 2 MB	NPU
WideResNet50	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	1.86 ms	0 - 2 MB	NPU
WideResNet50	QNN_DLC	w8a8	Qualcomm® SA8775P	2.017 ms	0 - 54 MB	NPU
WideResNet50	QNN_DLC	w8a8	Qualcomm® SA8775P	2.017 ms	0 - 54 MB	NPU
WideResNet50	QNN_DLC	w8a8	Qualcomm® SA8775P	2.017 ms	0 - 54 MB	NPU
WideResNet50	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	2.864 ms	0 - 209 MB	NPU
WideResNet50	QNN_DLC	w8a8	Qualcomm® QCM6690	18.583 ms	0 - 211 MB	NPU
WideResNet50	QNN_DLC	w8a8	Qualcomm® QCS9075	1.904 ms	0 - 2 MB	NPU
WideResNet50	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	1.162 ms	0 - 53 MB	NPU
WideResNet50	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	2.526 ms	0 - 127 MB	NPU
WideResNet50	QNN_DLC	w8a8	Qualcomm® SA7255P	3.973 ms	0 - 53 MB	NPU
WideResNet50	QNN_DLC	w8a8	Qualcomm® SA8295P	2.743 ms	0 - 51 MB	NPU
WideResNet50	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	2.864 ms	0 - 209 MB	NPU
WideResNet50	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	2.525 ms	0 - 102 MB	NPU
WideResNet50	TFLITE	float	Snapdragon® 8 Elite Mobile	2.855 ms	0 - 106 MB	NPU
WideResNet50	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	3.543 ms	0 - 213 MB	NPU
WideResNet50	TFLITE	float	Qualcomm® QCS8550 (Proxy)	4.743 ms	0 - 2 MB	NPU
WideResNet50	TFLITE	float	Qualcomm® SA8775P	7.04 ms	0 - 101 MB	NPU
WideResNet50	TFLITE	float	Qualcomm® SA8775P	7.04 ms	0 - 101 MB	NPU
WideResNet50	TFLITE	float	Qualcomm® SA8775P	7.04 ms	0 - 101 MB	NPU
WideResNet50	TFLITE	float	Qualcomm® SA7255P	23.77 ms	0 - 100 MB	NPU
WideResNet50	TFLITE	float	Qualcomm® QCS8450 (Proxy)	8.183 ms	0 - 189 MB	NPU
WideResNet50	TFLITE	float	Qualcomm® SA8295P	7.744 ms	0 - 88 MB	NPU
WideResNet50	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	2.855 ms	0 - 106 MB	NPU
WideResNet50	TFLITE	float	Qualcomm® QCS9075	7.051 ms	0 - 134 MB	NPU
WideResNet50	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	1.046 ms	0 - 56 MB	NPU
WideResNet50	TFLITE	w8a8	Snapdragon® 8 Elite Mobile	1.126 ms	0 - 58 MB	NPU
WideResNet50	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	1.347 ms	0 - 127 MB	NPU
WideResNet50	TFLITE	w8a8	Qualcomm® QCS6490	6.975 ms	0 - 68 MB	NPU
WideResNet50	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	1.732 ms	0 - 3 MB	NPU
WideResNet50	TFLITE	w8a8	Qualcomm® SA8775P	1.898 ms	0 - 53 MB	NPU
WideResNet50	TFLITE	w8a8	Qualcomm® SA8775P	1.898 ms	0 - 53 MB	NPU
WideResNet50	TFLITE	w8a8	Qualcomm® SA8775P	1.898 ms	0 - 53 MB	NPU
WideResNet50	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	2.722 ms	0 - 207 MB	NPU
WideResNet50	TFLITE	w8a8	Qualcomm® QCM6690	17.106 ms	0 - 213 MB	NPU
WideResNet50	TFLITE	w8a8	Qualcomm® QCS9075	1.739 ms	0 - 68 MB	NPU
WideResNet50	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	1.126 ms	0 - 58 MB	NPU
WideResNet50	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	2.39 ms	0 - 131 MB	NPU
WideResNet50	TFLITE	w8a8	Qualcomm® SA7255P	3.762 ms	0 - 52 MB	NPU
WideResNet50	TFLITE	w8a8	Qualcomm® SA8295P	2.58 ms	0 - 50 MB	NPU
WideResNet50	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	2.722 ms	0 - 207 MB	NPU

License

• The license for the original implementation of WideResNet50 can be found here.

References

• Wide Residual 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.