MNASNet05: Optimized for Qualcomm Devices

MNASNet05 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 MNASNet05 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 MNASNet05 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 MNASNet05 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: 2.21M
• Model size (float): 8.45 MB
• Model size (w8a16): 2.79 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
MNASNet05	ONNX	float	Snapdragon® X2 Elite	0.225 ms	212 - 212 MB	NPU
MNASNet05	ONNX	float	Snapdragon® X Elite	0.484 ms	181 - 181 MB	NPU
MNASNet05	ONNX	float	Snapdragon® 8 Gen 3 Mobile	0.33 ms	0 - 35 MB	NPU
MNASNet05	ONNX	float	Snapdragon® 8 Gen 1 Mobile	0.877 ms	1 - 44 MB	NPU
MNASNet05	ONNX	float	Qualcomm® QCS8550 (Proxy)	0.484 ms	0 - 69 MB	NPU
MNASNet05	ONNX	float	Qualcomm® QCS8450	0.877 ms	1 - 44 MB	NPU
MNASNet05	ONNX	float	Snapdragon® 8 Elite Mobile	0.268 ms	0 - 28 MB	NPU
MNASNet05	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.222 ms	1 - 24 MB	NPU
MNASNet05	ONNX	float	Qualcomm® QCS9075	0.765 ms	0 - 50 MB	NPU
MNASNet05	ONNX	float	Qualcomm® QCS8750	0.268 ms	0 - 28 MB	NPU
MNASNet05	ONNX	float	Qualcomm® QCS7181	0.484 ms	181 - 181 MB	NPU
MNASNet05	ONNX	w8a16	Snapdragon® X2 Elite	0.214 ms	213 - 213 MB	NPU
MNASNet05	ONNX	w8a16	Snapdragon® X Elite	0.513 ms	149 - 149 MB	NPU
MNASNet05	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	0.347 ms	0 - 32 MB	NPU
MNASNet05	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	0.679 ms	0 - 35 MB	NPU
MNASNet05	ONNX	w8a16	Qualcomm® QCS6490	1.686 ms	0 - 50 MB	NPU
MNASNet05	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	0.516 ms	0 - 15 MB	NPU
MNASNet05	ONNX	w8a16	Qualcomm® QCS8450	0.679 ms	0 - 35 MB	NPU
MNASNet05	ONNX	w8a16	Qualcomm® QCS9075	0.694 ms	0 - 48 MB	NPU
MNASNet05	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	0.22 ms	0 - 28 MB	NPU
MNASNet05	ONNX	w8a16	Snapdragon® 8 Elite Mobile	0.266 ms	0 - 30 MB	NPU
MNASNet05	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	0.537 ms	0 - 27 MB	NPU
MNASNet05	ONNX	w8a16	Qualcomm® QCM6690	2.289 ms	0 - 141 MB	NPU
MNASNet05	ONNX	w8a16	Qualcomm® QCS7790	0.537 ms	0 - 27 MB	NPU
MNASNet05	ONNX	w8a16	Qualcomm® QCS8750	0.266 ms	0 - 30 MB	NPU
MNASNet05	ONNX	w8a16	Qualcomm® QCS7181	0.513 ms	149 - 149 MB	NPU
MNASNet05	QNN_DLC	float	Snapdragon® X2 Elite	0.406 ms	1 - 1 MB	NPU
MNASNet05	QNN_DLC	float	Snapdragon® X Elite	0.919 ms	1 - 1 MB	NPU
MNASNet05	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	0.519 ms	0 - 44 MB	NPU
MNASNet05	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	1.569 ms	0 - 52 MB	NPU
MNASNet05	QNN_DLC	float	Qualcomm® QCS8275	2.32 ms	1 - 27 MB	NPU
MNASNet05	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	0.79 ms	1 - 2 MB	NPU
MNASNet05	QNN_DLC	float	Qualcomm® QCS8450	1.569 ms	0 - 52 MB	NPU
MNASNet05	QNN_DLC	float	Snapdragon® 8 Elite Mobile	0.384 ms	1 - 28 MB	NPU
MNASNet05	QNN_DLC	float	Qualcomm® SA8295P	1.435 ms	0 - 27 MB	NPU
MNASNet05	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.295 ms	1 - 29 MB	NPU
MNASNet05	QNN_DLC	float	Qualcomm® SA7255P	2.32 ms	1 - 27 MB	NPU
MNASNet05	QNN_DLC	float	Qualcomm® QCS9075	0.975 ms	3 - 5 MB	NPU
MNASNet05	QNN_DLC	float	Qualcomm® QCS8750	0.384 ms	1 - 28 MB	NPU
MNASNet05	QNN_DLC	float	Qualcomm® QCS7181	0.919 ms	1 - 1 MB	NPU
MNASNet05	QNN_DLC	w8a16	Snapdragon® X2 Elite	0.399 ms	0 - 0 MB	NPU
MNASNet05	QNN_DLC	w8a16	Snapdragon® X Elite	0.891 ms	0 - 0 MB	NPU
MNASNet05	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	0.524 ms	0 - 38 MB	NPU
MNASNet05	QNN_DLC	w8a16	Snapdragon® 8 Gen 1 Mobile	0.953 ms	0 - 45 MB	NPU
MNASNet05	QNN_DLC	w8a16	Qualcomm® QCS6490	2.25 ms	2 - 4 MB	NPU
MNASNet05	QNN_DLC	w8a16	Qualcomm® QCS8275	1.721 ms	0 - 28 MB	NPU
MNASNet05	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	0.762 ms	0 - 19 MB	NPU
MNASNet05	QNN_DLC	w8a16	Qualcomm® QCS8450	0.953 ms	0 - 45 MB	NPU
MNASNet05	QNN_DLC	w8a16	Qualcomm® QCS9075	0.917 ms	2 - 4 MB	NPU
MNASNet05	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	0.305 ms	0 - 28 MB	NPU
MNASNet05	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	0.352 ms	0 - 28 MB	NPU
MNASNet05	QNN_DLC	w8a16	Qualcomm® SA8295P	1.219 ms	0 - 25 MB	NPU
MNASNet05	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	0.79 ms	0 - 27 MB	NPU
MNASNet05	QNN_DLC	w8a16	Qualcomm® QCM6690	3.052 ms	0 - 140 MB	NPU
MNASNet05	QNN_DLC	w8a16	Qualcomm® SA7255P	1.721 ms	0 - 28 MB	NPU
MNASNet05	QNN_DLC	w8a16	Qualcomm® QCS7790	0.79 ms	0 - 27 MB	NPU
MNASNet05	QNN_DLC	w8a16	Qualcomm® QCS8750	0.352 ms	0 - 28 MB	NPU
MNASNet05	QNN_DLC	w8a16	Qualcomm® QCS7181	0.891 ms	0 - 0 MB	NPU
MNASNet05	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	0.517 ms	0 - 45 MB	NPU
MNASNet05	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	1.575 ms	0 - 47 MB	NPU
MNASNet05	TFLITE	float	Qualcomm® QCS8275	2.335 ms	0 - 28 MB	NPU
MNASNet05	TFLITE	float	Qualcomm® QCS8550 (Proxy)	0.799 ms	0 - 2 MB	NPU
MNASNet05	TFLITE	float	Qualcomm® SA8775P	2.115 ms	0 - 30 MB	GPU
MNASNet05	TFLITE	float	Qualcomm® SA8650P	2.115 ms	0 - 30 MB	GPU
MNASNet05	TFLITE	float	Qualcomm® SA8255P	2.115 ms	0 - 30 MB	GPU
MNASNet05	TFLITE	float	Qualcomm® QCS8450	1.575 ms	0 - 47 MB	NPU
MNASNet05	TFLITE	float	Snapdragon® 8 Elite Mobile	0.381 ms	0 - 29 MB	NPU
MNASNet05	TFLITE	float	Qualcomm® SA8295P	1.449 ms	0 - 28 MB	NPU
MNASNet05	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.295 ms	0 - 30 MB	NPU
MNASNet05	TFLITE	float	Qualcomm® SA7255P	2.335 ms	0 - 28 MB	NPU
MNASNet05	TFLITE	float	Qualcomm® QCS9075	0.985 ms	0 - 8 MB	NPU
MNASNet05	TFLITE	float	Qualcomm® QCS8750	0.381 ms	0 - 29 MB	NPU

License

• The license for the original implementation of MNASNet05 can be found here.

References

• MnasNet: Platform-Aware Neural Architecture Search for Mobile
• 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.