NASNet: Optimized for Qualcomm Devices

NASNet is a CNN-based architecture discovered via Neural Architecture Search (NAS) that can classify images from the Imagenet dataset.

This is based on the implementation of NASNet 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 1.27.1	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit NASNet 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 NASNet on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_classification

Model Stats:

• Model checkpoint: nasnetalarge.tf_in1k
• Input resolution: 224x224
• GMACs: 5.9
• Activations (M): 19.4
• Number of parameters: 88.7M
• Model size (float): 338 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
NASNet	ONNX	float	Snapdragon® X2 Elite	9.635 ms	1 - 1 MB	NPU
NASNet	ONNX	float	Snapdragon® X Elite	18.905 ms	189 - 189 MB	NPU
NASNet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	14.121 ms	1 - 571 MB	NPU
NASNet	ONNX	float	Snapdragon® 8 Gen 1 Mobile	52.409 ms	2 - 519 MB	NPU
NASNet	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	18.849 ms	0 - 194 MB	NPU
NASNet	ONNX	float	Qualcomm® QCS8450	52.409 ms	2 - 519 MB	NPU
NASNet	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	27.065 ms	1 - 5 MB	NPU
NASNet	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	18.905 ms	189 - 189 MB	NPU
NASNet	ONNX	float	Qualcomm® Dragonwing™ Q-8750	11.649 ms	1 - 405 MB	NPU
NASNet	ONNX	float	Snapdragon® 8 Elite Mobile	11.649 ms	1 - 405 MB	NPU
NASNet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	9.506 ms	0 - 415 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® X2 Elite	8.868 ms	1 - 1 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® X Elite	16.828 ms	1 - 1 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	12.314 ms	0 - 516 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	48.435 ms	0 - 475 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8275	89.202 ms	2 - 351 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	16.219 ms	1 - 4 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® SA8775P	24.393 ms	1 - 354 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® SA8650P	24.393 ms	1 - 354 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® SA8255P	24.393 ms	1 - 354 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® QCS8450	48.435 ms	0 - 475 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	24.257 ms	1 - 4 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	16.828 ms	1 - 1 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	9.824 ms	1 - 352 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® SA7255P	89.202 ms	2 - 351 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® SA8295P	38.659 ms	1 - 310 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® 8 Elite Mobile	9.824 ms	1 - 352 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	7.816 ms	0 - 370 MB	NPU
NASNet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	12.71 ms	0 - 695 MB	NPU
NASNet	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	49.651 ms	0 - 653 MB	NPU
NASNet	TFLITE	float	Qualcomm® Dragonwing™ QCS8275	89.633 ms	0 - 539 MB	NPU
NASNet	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	16.666 ms	0 - 3 MB	NPU
NASNet	TFLITE	float	Qualcomm® SA8775P	24.747 ms	0 - 537 MB	NPU
NASNet	TFLITE	float	Qualcomm® SA8650P	24.747 ms	0 - 537 MB	NPU
NASNet	TFLITE	float	Qualcomm® SA8255P	24.747 ms	0 - 537 MB	NPU
NASNet	TFLITE	float	Qualcomm® QCS8450	49.651 ms	0 - 653 MB	NPU
NASNet	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	25.177 ms	0 - 192 MB	NPU
NASNet	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	10.295 ms	0 - 539 MB	NPU
NASNet	TFLITE	float	Qualcomm® SA7255P	89.633 ms	0 - 539 MB	NPU
NASNet	TFLITE	float	Qualcomm® SA8295P	40.084 ms	0 - 486 MB	NPU
NASNet	TFLITE	float	Snapdragon® 8 Elite Mobile	10.295 ms	0 - 539 MB	NPU
NASNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	8.229 ms	0 - 549 MB	NPU

License

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

References

• Learning Transferable Architectures for Scalable 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.