ResNet18: Optimized for Qualcomm Devices

ResNet18 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 ResNet18 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 1.24.1	Download
ONNX	w8a8	Universal	QAIRT 2.42, ONNX Runtime 1.24.1	Download
QNN_DLC	float	Universal	QAIRT 2.43	Download
QNN_DLC	w8a8	Universal	QAIRT 2.43	Download
TFLITE	float	Universal	QAIRT 2.43, TFLite 2.17.0	Download
TFLITE	w8a8	Universal	QAIRT 2.43, TFLite 2.17.0	Download

For more device-specific assets and performance metrics, visit ResNet18 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 ResNet18 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: 11.7M
• Model size (float): 44.6 MB
• Model size (w8a8): 11.3 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
ResNet18	ONNX	float	Snapdragon® X Elite	1.171 ms	22 - 22 MB	NPU
ResNet18	ONNX	float	Snapdragon® 8 Gen 3 Mobile	0.778 ms	0 - 35 MB	NPU
ResNet18	ONNX	float	Qualcomm® QCS8550 (Proxy)	1.072 ms	0 - 43 MB	NPU
ResNet18	ONNX	float	Qualcomm® QCS9075	1.796 ms	1 - 3 MB	NPU
ResNet18	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	0.638 ms	0 - 26 MB	NPU
ResNet18	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.557 ms	1 - 26 MB	NPU
ResNet18	ONNX	float	Snapdragon® X2 Elite	0.524 ms	23 - 23 MB	NPU
ResNet18	ONNX	w8a8	Snapdragon® X Elite	0.637 ms	11 - 11 MB	NPU
ResNet18	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	0.417 ms	0 - 47 MB	NPU
ResNet18	ONNX	w8a8	Qualcomm® QCS6490	13.584 ms	6 - 21 MB	CPU
ResNet18	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	0.557 ms	0 - 2 MB	NPU
ResNet18	ONNX	w8a8	Qualcomm® QCS9075	0.645 ms	0 - 3 MB	NPU
ResNet18	ONNX	w8a8	Qualcomm® QCM6690	11.393 ms	8 - 14 MB	CPU
ResNet18	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	0.355 ms	0 - 22 MB	NPU
ResNet18	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	8.74 ms	6 - 12 MB	CPU
ResNet18	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.347 ms	0 - 27 MB	NPU
ResNet18	ONNX	w8a8	Snapdragon® X2 Elite	0.268 ms	11 - 11 MB	NPU
ResNet18	QNN_DLC	float	Snapdragon® X Elite	1.474 ms	1 - 1 MB	NPU
ResNet18	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	0.916 ms	0 - 38 MB	NPU
ResNet18	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	5.799 ms	1 - 24 MB	NPU
ResNet18	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	1.337 ms	1 - 2 MB	NPU
ResNet18	QNN_DLC	float	Qualcomm® SA8775P	1.971 ms	0 - 26 MB	NPU
ResNet18	QNN_DLC	float	Qualcomm® QCS9075	2.065 ms	3 - 5 MB	NPU
ResNet18	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	2.449 ms	0 - 37 MB	NPU
ResNet18	QNN_DLC	float	Qualcomm® SA7255P	5.799 ms	1 - 24 MB	NPU
ResNet18	QNN_DLC	float	Qualcomm® SA8295P	2.312 ms	0 - 20 MB	NPU
ResNet18	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	0.711 ms	0 - 27 MB	NPU
ResNet18	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.589 ms	1 - 28 MB	NPU
ResNet18	QNN_DLC	float	Snapdragon® X2 Elite	0.68 ms	1 - 1 MB	NPU
ResNet18	QNN_DLC	w8a8	Snapdragon® X Elite	0.604 ms	0 - 0 MB	NPU
ResNet18	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	0.387 ms	0 - 47 MB	NPU
ResNet18	QNN_DLC	w8a8	Qualcomm® QCS6490	1.805 ms	0 - 2 MB	NPU
ResNet18	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	1.271 ms	0 - 24 MB	NPU
ResNet18	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	0.518 ms	0 - 1 MB	NPU
ResNet18	QNN_DLC	w8a8	Qualcomm® SA8775P	0.707 ms	0 - 24 MB	NPU
ResNet18	QNN_DLC	w8a8	Qualcomm® QCS9075	0.624 ms	2 - 4 MB	NPU
ResNet18	QNN_DLC	w8a8	Qualcomm® QCM6690	3.245 ms	0 - 31 MB	NPU
ResNet18	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	0.749 ms	0 - 49 MB	NPU
ResNet18	QNN_DLC	w8a8	Qualcomm® SA7255P	1.271 ms	0 - 24 MB	NPU
ResNet18	QNN_DLC	w8a8	Qualcomm® SA8295P	0.911 ms	0 - 21 MB	NPU
ResNet18	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	0.3 ms	0 - 27 MB	NPU
ResNet18	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	0.683 ms	0 - 31 MB	NPU
ResNet18	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.258 ms	0 - 25 MB	NPU
ResNet18	QNN_DLC	w8a8	Snapdragon® X2 Elite	0.322 ms	0 - 0 MB	NPU
ResNet18	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	0.92 ms	0 - 63 MB	NPU
ResNet18	TFLITE	float	Qualcomm® QCS8275 (Proxy)	5.743 ms	0 - 27 MB	NPU
ResNet18	TFLITE	float	Qualcomm® QCS8550 (Proxy)	1.333 ms	0 - 3 MB	NPU
ResNet18	TFLITE	float	Qualcomm® SA8775P	1.965 ms	0 - 30 MB	NPU
ResNet18	TFLITE	float	Qualcomm® QCS9075	2.031 ms	0 - 25 MB	NPU
ResNet18	TFLITE	float	Qualcomm® QCS8450 (Proxy)	2.453 ms	0 - 61 MB	NPU
ResNet18	TFLITE	float	Qualcomm® SA7255P	5.743 ms	0 - 27 MB	NPU
ResNet18	TFLITE	float	Qualcomm® SA8295P	2.273 ms	0 - 22 MB	NPU
ResNet18	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	0.707 ms	0 - 26 MB	NPU
ResNet18	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.588 ms	0 - 30 MB	NPU
ResNet18	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	0.287 ms	0 - 46 MB	NPU
ResNet18	TFLITE	w8a8	Qualcomm® QCS6490	1.458 ms	0 - 13 MB	NPU
ResNet18	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	1.03 ms	0 - 23 MB	NPU
ResNet18	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	0.388 ms	0 - 1 MB	NPU
ResNet18	TFLITE	w8a8	Qualcomm® SA8775P	0.572 ms	0 - 25 MB	NPU
ResNet18	TFLITE	w8a8	Qualcomm® QCS9075	0.475 ms	0 - 13 MB	NPU
ResNet18	TFLITE	w8a8	Qualcomm® QCM6690	2.801 ms	0 - 30 MB	NPU
ResNet18	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	0.615 ms	0 - 48 MB	NPU
ResNet18	TFLITE	w8a8	Qualcomm® SA7255P	1.03 ms	0 - 23 MB	NPU
ResNet18	TFLITE	w8a8	Qualcomm® SA8295P	0.765 ms	0 - 21 MB	NPU
ResNet18	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	0.235 ms	0 - 26 MB	NPU
ResNet18	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	0.55 ms	0 - 30 MB	NPU
ResNet18	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.224 ms	0 - 25 MB	NPU

License

• The license for the original implementation of ResNet18 can be found here.

References

• Deep Residual Learning for Image Recognition
• 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.