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 2.50, ONNX Runtime 1.30.0	Download
QNN_DLC	float	Universal	QAIRT 2.50	Download
TFLITE	float	Universal	QAIRT 2.50	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:

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

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
NASNet	ONNX	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	8.792 ms	0 - 361 MB	NPU
NASNet	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	10.823 ms	0 - 352 MB	NPU
NASNet	ONNX	float	Snapdragon® X2 Elite	9.034 ms	1 - 1 MB	NPU
NASNet	ONNX	float	Snapdragon® X Elite	17.216 ms	188 - 188 MB	NPU
NASNet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	13.174 ms	0 - 515 MB	NPU
NASNet	ONNX	float	Snapdragon® 8 Gen 1 Mobile	39.327 ms	1 - 484 MB	NPU
NASNet	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	28.693 ms	1 - 7 MB	NPU
NASNet	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	17.305 ms	0 - 193 MB	NPU
NASNet	ONNX	float	Qualcomm® QCS8450	39.327 ms	1 - 484 MB	NPU
NASNet	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	25.199 ms	1 - 6 MB	NPU
NASNet	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	17.216 ms	188 - 188 MB	NPU
NASNet	ONNX	float	Qualcomm® Dragonwing™ Q-8750	10.823 ms	0 - 352 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	7.811 ms	0 - 332 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	9.497 ms	1 - 323 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® X2 Elite	8.511 ms	1 - 1 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® X Elite	16.07 ms	1 - 1 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	11.621 ms	0 - 479 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	35.678 ms	0 - 450 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	26.745 ms	1 - 5 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	15.862 ms	1 - 571 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® SA8775P	23.627 ms	1 - 321 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® SA8650P	23.627 ms	1 - 321 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® SA8255P	23.627 ms	1 - 321 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® QCS8450	35.678 ms	0 - 450 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	24.142 ms	1 - 5 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	16.07 ms	1 - 1 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	9.497 ms	1 - 323 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® SA7255P	87.672 ms	1 - 321 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® SA8295P	38.633 ms	1 - 290 MB	NPU
NASNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	8.153 ms	0 - 524 MB	NPU
NASNet	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	9.928 ms	0 - 508 MB	NPU
NASNet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	12.297 ms	0 - 670 MB	NPU
NASNet	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	35.996 ms	0 - 636 MB	NPU
NASNet	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	27.242 ms	0 - 193 MB	NPU
NASNet	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	16.173 ms	0 - 3 MB	NPU
NASNet	TFLITE	float	Qualcomm® QCS8450	35.996 ms	0 - 636 MB	NPU
NASNet	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	24.512 ms	0 - 192 MB	NPU
NASNet	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	9.928 ms	0 - 508 MB	NPU
NASNet	TFLITE	float	Qualcomm® SA8295P	39.877 ms	0 - 468 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.