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 2.42, ONNX Runtime 1.24.3	Download
ONNX	w8a8	Universal	QAIRT 2.42, ONNX Runtime 1.24.3	Download
QNN_DLC	float	Universal	QAIRT 2.43	Download
QNN_DLC	w8a8	Universal	QAIRT 2.43	Download
TFLITE	float	Universal	QAIRT 2.43, TFLite 2.19.1	Download
TFLITE	w8a8	Universal	QAIRT 2.43, TFLite 2.19.1	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® 8 Elite Gen 5 Mobile	0.832 ms	1 - 88 MB	NPU
RegNet	ONNX	float	Snapdragon® X2 Elite	0.905 ms	39 - 39 MB	NPU
RegNet	ONNX	float	Snapdragon® X Elite	1.979 ms	39 - 39 MB	NPU
RegNet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	1.244 ms	0 - 135 MB	NPU
RegNet	ONNX	float	Qualcomm® QCS8550 (Proxy)	1.754 ms	1 - 4 MB	NPU
RegNet	ONNX	float	Qualcomm® QCS9075	2.856 ms	0 - 4 MB	NPU
RegNet	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	0.997 ms	0 - 87 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.505 ms	0 - 89 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® X2 Elite	0.483 ms	20 - 20 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® X Elite	1.116 ms	20 - 20 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	0.647 ms	0 - 135 MB	NPU
RegNet	ONNX	w8a8	Qualcomm® QCS6490	27.892 ms	7 - 19 MB	CPU
RegNet	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	0.902 ms	0 - 28 MB	NPU
RegNet	ONNX	w8a8	Qualcomm® QCS9075	1.079 ms	0 - 3 MB	NPU
RegNet	ONNX	w8a8	Qualcomm® QCM6690	18.24 ms	9 - 18 MB	CPU
RegNet	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	0.547 ms	0 - 79 MB	NPU
RegNet	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	13.774 ms	9 - 18 MB	CPU
RegNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.896 ms	1 - 81 MB	NPU
RegNet	QNN_DLC	float	Snapdragon® X2 Elite	1.272 ms	1 - 1 MB	NPU
RegNet	QNN_DLC	float	Snapdragon® X Elite	2.313 ms	1 - 1 MB	NPU
RegNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	1.39 ms	0 - 128 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	9.913 ms	1 - 77 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	2.087 ms	1 - 2 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® SA8775P	3.316 ms	0 - 78 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® QCS9075	3.064 ms	1 - 3 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	3.473 ms	0 - 115 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® SA7255P	9.913 ms	1 - 77 MB	NPU
RegNet	QNN_DLC	float	Qualcomm® SA8295P	3.513 ms	0 - 65 MB	NPU
RegNet	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	1.142 ms	0 - 79 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.445 ms	0 - 80 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® X2 Elite	0.592 ms	0 - 0 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® X Elite	1.064 ms	0 - 0 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	0.631 ms	0 - 107 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS6490	2.701 ms	0 - 2 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	2.288 ms	0 - 76 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	0.881 ms	0 - 88 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® SA8775P	1.299 ms	0 - 78 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS9075	1.081 ms	0 - 2 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCM6690	7.096 ms	0 - 196 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	1.299 ms	0 - 111 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® SA7255P	2.288 ms	0 - 76 MB	NPU
RegNet	QNN_DLC	w8a8	Qualcomm® SA8295P	1.632 ms	0 - 75 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	0.488 ms	0 - 79 MB	NPU
RegNet	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	1.173 ms	0 - 77 MB	NPU
RegNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.899 ms	0 - 104 MB	NPU
RegNet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	1.397 ms	0 - 159 MB	NPU
RegNet	TFLITE	float	Qualcomm® QCS8275 (Proxy)	9.9 ms	0 - 101 MB	NPU
RegNet	TFLITE	float	Qualcomm® QCS8550 (Proxy)	2.068 ms	0 - 2 MB	NPU
RegNet	TFLITE	float	Qualcomm® SA8775P	3.342 ms	0 - 102 MB	NPU
RegNet	TFLITE	float	Qualcomm® QCS9075	3.053 ms	0 - 42 MB	NPU
RegNet	TFLITE	float	Qualcomm® QCS8450 (Proxy)	3.449 ms	0 - 143 MB	NPU
RegNet	TFLITE	float	Qualcomm® SA7255P	9.9 ms	0 - 101 MB	NPU
RegNet	TFLITE	float	Qualcomm® SA8295P	3.48 ms	0 - 83 MB	NPU
RegNet	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	1.107 ms	0 - 95 MB	NPU
RegNet	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.389 ms	0 - 79 MB	NPU
RegNet	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	0.523 ms	0 - 118 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCS6490	2.374 ms	0 - 21 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	1.978 ms	0 - 76 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	0.727 ms	0 - 126 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® SA8775P	1.144 ms	0 - 77 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCS9075	0.893 ms	0 - 22 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCM6690	6.606 ms	0 - 191 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	1.097 ms	0 - 111 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® SA7255P	1.978 ms	0 - 76 MB	NPU
RegNet	TFLITE	w8a8	Qualcomm® SA8295P	1.419 ms	0 - 72 MB	NPU
RegNet	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	0.424 ms	0 - 70 MB	NPU
RegNet	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	0.987 ms	0 - 73 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.