RegNet: Optimized for Qualcomm Devices

RegNet 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 RegNet 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 RegNet 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 RegNet 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: 15.3M
• Model size (float): 58.3 MB
• Model size (w8a8): 15.4 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
RegNet	ONNX	float	Snapdragon® X2 Elite	0.904 ms	212 - 212 MB	NPU
RegNet	ONNX	float	Snapdragon® X Elite	1.755 ms	148 - 148 MB	NPU
RegNet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	1.212 ms	1 - 117 MB	NPU
RegNet	ONNX	float	Snapdragon® 8 Gen 1 Mobile	2.72 ms	1 - 108 MB	NPU
RegNet	ONNX	float	Qualcomm® QCS8550 (Proxy)	1.789 ms	0 - 43 MB	NPU
RegNet	ONNX	float	Qualcomm® QCS8450	2.72 ms	1 - 108 MB	NPU
RegNet	ONNX	float	Snapdragon® 8 Elite Mobile	0.995 ms	0 - 77 MB	NPU
RegNet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.838 ms	1 - 76 MB	NPU
RegNet	ONNX	float	Qualcomm® QCS9075	2.835 ms	1 - 46 MB	NPU
RegNet	ONNX	float	Qualcomm® QCS8750	0.995 ms	0 - 77 MB	NPU
RegNet	ONNX	float	Qualcomm® QCS7181	1.755 ms	148 - 148 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® X2 Elite	0.469 ms	213 - 213 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® X Elite	0.936 ms	149 - 149 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	0.656 ms	0 - 126 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® 8 Gen 1 Mobile	1.289 ms	0 - 124 MB	NPU
RegNet	ONNX	w8a8	Qualcomm® QCS6490	2.37 ms	0 - 45 MB	NPU
RegNet	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	0.891 ms	0 - 3 MB	NPU
RegNet	ONNX	w8a8	Qualcomm® QCS8450	1.289 ms	0 - 124 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.508 ms	0 - 84 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	1.154 ms	0 - 82 MB	NPU
RegNet	ONNX	w8a8	Qualcomm® QCM6690	7.237 ms	0 - 201 MB	NPU
RegNet	ONNX	w8a8	Qualcomm® QCS9075	1.064 ms	0 - 45 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® 8 Elite Mobile	0.55 ms	0 - 90 MB	NPU
RegNet	ONNX	w8a8	Qualcomm® QCS7790	1.154 ms	0 - 82 MB	NPU
RegNet	ONNX	w8a8	Qualcomm® QCS8750	0.55 ms	0 - 90 MB	NPU
RegNet	ONNX	w8a8	Qualcomm® QCS7181	0.936 ms	149 - 149 MB	NPU
RegNet	QNN_DLC	float	Snapdragon® X2 Elite	1.228 ms	1 - 1 MB	NPU
RegNet	QNN_DLC	float	Snapdragon® X Elite	2.247 ms	1 - 1 MB	NPU
RegNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	1.401 ms	0 - 125 MB	NPU
RegNet	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	3.502 ms	1 - 117 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® QCS8275	9.863 ms	1 - 72 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	2.042 ms	1 - 124 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® QCS8450	3.502 ms	1 - 117 MB	NPU
RegNet	QNN_DLC	float	Snapdragon® 8 Elite Mobile	1.107 ms	1 - 74 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® SA8295P	3.471 ms	0 - 61 MB	NPU
RegNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.902 ms	1 - 77 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® SA7255P	9.863 ms	1 - 72 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® QCS9075	3.034 ms	3 - 5 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® QCS8750	1.107 ms	1 - 74 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® QCS7181	2.247 ms	1 - 1 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® X2 Elite	0.58 ms	0 - 0 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® X Elite	1.053 ms	0 - 0 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	0.637 ms	0 - 111 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® 8 Gen 1 Mobile	1.307 ms	0 - 112 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS6490	2.668 ms	2 - 4 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS8275	2.283 ms	0 - 80 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	0.866 ms	0 - 4 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS8450	1.307 ms	0 - 112 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.443 ms	0 - 81 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	1.174 ms	0 - 80 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCM6690	7.122 ms	0 - 199 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS9075	1.054 ms	0 - 2 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® SA7255P	2.283 ms	0 - 80 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® 8 Elite Mobile	0.482 ms	0 - 78 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® SA8295P	1.605 ms	0 - 78 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS7790	1.174 ms	0 - 80 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS8750	0.482 ms	0 - 78 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS7181	1.053 ms	0 - 0 MB	NPU
RegNet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	1.401 ms	0 - 153 MB	NPU
RegNet	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	3.47 ms	0 - 138 MB	NPU
RegNet	TFLITE	float	Qualcomm® QCS8275	9.855 ms	0 - 89 MB	NPU
RegNet	TFLITE	float	Qualcomm® QCS8550 (Proxy)	2.038 ms	0 - 2 MB	NPU
RegNet	TFLITE	float	Qualcomm® SA8775P	14.573 ms	0 - 29 MB	GPU
RegNet	TFLITE	float	Qualcomm® SA8650P	14.573 ms	0 - 29 MB	GPU
RegNet	TFLITE	float	Qualcomm® SA8255P	14.573 ms	0 - 29 MB	GPU
RegNet	TFLITE	float	Qualcomm® QCS8450	3.47 ms	0 - 138 MB	NPU
RegNet	TFLITE	float	Snapdragon® 8 Elite Mobile	1.104 ms	0 - 92 MB	NPU
RegNet	TFLITE	float	Qualcomm® SA8295P	3.48 ms	0 - 72 MB	NPU
RegNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.899 ms	0 - 86 MB	NPU
RegNet	TFLITE	float	Qualcomm® SA7255P	9.855 ms	0 - 89 MB	NPU
RegNet	TFLITE	float	Qualcomm® QCS9075	3.03 ms	0 - 42 MB	NPU
RegNet	TFLITE	float	Qualcomm® QCS8750	1.104 ms	0 - 92 MB	NPU
RegNet	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	0.527 ms	0 - 120 MB	NPU
RegNet	TFLITE	w8a8	Snapdragon® 8 Gen 1 Mobile	1.105 ms	0 - 116 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCS6490	2.368 ms	0 - 22 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCS8275	2.004 ms	0 - 75 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	0.707 ms	0 - 2 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® SA8775P	15.165 ms	1 - 31 MB	GPU
RegNet	TFLITE	w8a8	Qualcomm® SA8650P	15.165 ms	1 - 31 MB	GPU
RegNet	TFLITE	w8a8	Qualcomm® SA8255P	15.165 ms	1 - 31 MB	GPU
RegNet	TFLITE	w8a8	Qualcomm® QCS8450	1.105 ms	0 - 116 MB	NPU
RegNet	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.393 ms	0 - 75 MB	NPU
RegNet	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	0.983 ms	0 - 73 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCM6690	6.647 ms	0 - 190 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCS9075	0.864 ms	0 - 22 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® SA7255P	2.004 ms	0 - 75 MB	NPU
RegNet	TFLITE	w8a8	Snapdragon® 8 Elite Mobile	0.421 ms	0 - 71 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® SA8295P	1.41 ms	0 - 72 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCS7790	0.983 ms	0 - 73 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCS8750	0.421 ms	0 - 71 MB	NPU

License

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

References

• Designing Network Design Spaces
• 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.