NASNet: Optimized for Qualcomm Devices

NASNet is a vision transformer model 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 2.42, ONNX Runtime 1.24.1	Download
ONNX	w8a8_mixed_fp16	Universal	QAIRT 2.42, ONNX Runtime 1.24.1	Download
QNN_DLC	float	Universal	QAIRT 2.43	Download
QNN_DLC	w8a8_mixed_fp16	Universal	QAIRT 2.43	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® 8 Elite Gen 5 Mobile	8.39 ms	1 - 673 MB	NPU
NASNet	ONNX	float	Snapdragon® X2 Elite	10.699 ms	189 - 189 MB	NPU
NASNet	ONNX	float	Snapdragon® X Elite	17.795 ms	188 - 188 MB	NPU
NASNet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	12.781 ms	1 - 838 MB	NPU
NASNet	ONNX	float	Qualcomm® QCS8550 (Proxy)	17.495 ms	0 - 197 MB	NPU
NASNet	ONNX	float	Qualcomm® QCS9075	28.261 ms	0 - 4 MB	NPU
NASNet	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	10.374 ms	0 - 640 MB	NPU
NASNet	ONNX	w8a8_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	4.516 ms	5 - 359 MB	NPU
NASNet	ONNX	w8a8_mixed_fp16	Snapdragon® X2 Elite	4.488 ms	100 - 100 MB	NPU
NASNet	ONNX	w8a8_mixed_fp16	Snapdragon® X Elite	12.004 ms	98 - 98 MB	NPU
NASNet	ONNX	w8a8_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	6.901 ms	6 - 478 MB	NPU
NASNet	ONNX	w8a8_mixed_fp16	Qualcomm® QCS8550 (Proxy)	9.747 ms	5 - 10 MB	NPU
NASNet	ONNX	w8a8_mixed_fp16	Qualcomm® QCS9075	11.55 ms	5 - 8 MB	NPU
NASNet	ONNX	w8a8_mixed_fp16	Snapdragon® 8 Elite For Galaxy Mobile	5.535 ms	5 - 361 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	8.769 ms	0 - 664 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® X2 Elite	10.154 ms	1 - 1 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® X Elite	19.289 ms	1 - 1 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	12.32 ms	0 - 813 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	54.013 ms	1 - 660 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	19.02 ms	1 - 3 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® QCS9075	28.662 ms	1 - 3 MB	NPU
NASNet	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	35.42 ms	0 - 793 MB	NPU
NASNet	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	11.17 ms	0 - 650 MB	NPU
NASNet	QNN_DLC	w8a8_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	3.843 ms	0 - 379 MB	NPU
NASNet	QNN_DLC	w8a8_mixed_fp16	Snapdragon® X2 Elite	4.111 ms	0 - 0 MB	NPU
NASNet	QNN_DLC	w8a8_mixed_fp16	Snapdragon® X Elite	9.146 ms	0 - 0 MB	NPU
NASNet	QNN_DLC	w8a8_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	6.018 ms	0 - 493 MB	NPU
NASNet	QNN_DLC	w8a8_mixed_fp16	Qualcomm® QCS8275 (Proxy)	16.35 ms	0 - 379 MB	NPU
NASNet	QNN_DLC	w8a8_mixed_fp16	Qualcomm® QCS8550 (Proxy)	8.843 ms	0 - 2 MB	NPU
NASNet	QNN_DLC	w8a8_mixed_fp16	Qualcomm® QCS9075	9.426 ms	0 - 2 MB	NPU
NASNet	QNN_DLC	w8a8_mixed_fp16	Qualcomm® QCS8450 (Proxy)	10.984 ms	0 - 505 MB	NPU
NASNet	QNN_DLC	w8a8_mixed_fp16	Snapdragon® 8 Elite For Galaxy Mobile	5.001 ms	0 - 378 MB	NPU
NASNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	5.624 ms	2 - 618 MB	NPU
NASNet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	8.785 ms	0 - 778 MB	NPU
NASNet	TFLITE	float	Qualcomm® QCS8275 (Proxy)	44.298 ms	0 - 629 MB	NPU
NASNet	TFLITE	float	Qualcomm® QCS8550 (Proxy)	12.486 ms	0 - 3 MB	NPU
NASNet	TFLITE	float	Qualcomm® QCS9075	15.553 ms	0 - 192 MB	NPU
NASNet	TFLITE	float	Qualcomm® QCS8450 (Proxy)	28.988 ms	0 - 758 MB	NPU
NASNet	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	6.952 ms	0 - 633 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.