GoogLeNet: Optimized for Qualcomm Devices

GoogLeNet 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 GoogLeNet 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	w8a8	Universal	QAIRT 2.45, ONNX Runtime 1.25.0	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	w8a8	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download
TFLITE	w8a8	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit GoogLeNet 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 GoogLeNet 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: 6.62M
• Model size (float): 25.3 MB
• Model size (w8a8): 6.54 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
GoogLeNet	ONNX	float	Snapdragon® X2 Elite	0.581 ms	212 - 212 MB	NPU
GoogLeNet	ONNX	float	Snapdragon® X Elite	0.864 ms	180 - 180 MB	NPU
GoogLeNet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	0.556 ms	1 - 37 MB	NPU
GoogLeNet	ONNX	float	Snapdragon® 8 Gen 1 Mobile	1.577 ms	1 - 42 MB	NPU
GoogLeNet	ONNX	float	Qualcomm® QCS8550 (Proxy)	0.852 ms	0 - 65 MB	NPU
GoogLeNet	ONNX	float	Qualcomm® QCS8450	1.577 ms	1 - 42 MB	NPU
GoogLeNet	ONNX	float	Snapdragon® 8 Elite Mobile	0.47 ms	0 - 31 MB	NPU
GoogLeNet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.39 ms	1 - 27 MB	NPU
GoogLeNet	ONNX	float	Qualcomm® QCS9075	1.584 ms	1 - 46 MB	NPU
GoogLeNet	ONNX	float	Qualcomm® QCS8750	0.47 ms	0 - 31 MB	NPU
GoogLeNet	ONNX	float	Qualcomm® QCS7181	0.864 ms	180 - 180 MB	NPU
GoogLeNet	ONNX	w8a8	Snapdragon® X2 Elite	0.161 ms	213 - 213 MB	NPU
GoogLeNet	ONNX	w8a8	Snapdragon® X Elite	0.293 ms	181 - 181 MB	NPU
GoogLeNet	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	0.228 ms	0 - 40 MB	NPU
GoogLeNet	ONNX	w8a8	Snapdragon® 8 Gen 1 Mobile	0.481 ms	0 - 45 MB	NPU
GoogLeNet	ONNX	w8a8	Qualcomm® QCS6490	1.149 ms	0 - 45 MB	NPU
GoogLeNet	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	0.309 ms	0 - 24 MB	NPU
GoogLeNet	ONNX	w8a8	Qualcomm® QCS8450	0.481 ms	0 - 45 MB	NPU
GoogLeNet	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.18 ms	0 - 34 MB	NPU
GoogLeNet	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	0.355 ms	0 - 32 MB	NPU
GoogLeNet	ONNX	w8a8	Qualcomm® QCM6690	2.343 ms	0 - 33 MB	NPU
GoogLeNet	ONNX	w8a8	Qualcomm® QCS9075	0.438 ms	0 - 52 MB	NPU
GoogLeNet	ONNX	w8a8	Snapdragon® 8 Elite Mobile	0.197 ms	0 - 34 MB	NPU
GoogLeNet	ONNX	w8a8	Qualcomm® QCS7790	0.355 ms	0 - 32 MB	NPU
GoogLeNet	ONNX	w8a8	Qualcomm® QCS8750	0.197 ms	0 - 34 MB	NPU
GoogLeNet	ONNX	w8a8	Qualcomm® QCS7181	0.293 ms	181 - 181 MB	NPU
GoogLeNet	QNN_DLC	float	Snapdragon® X2 Elite	0.52 ms	1 - 1 MB	NPU
GoogLeNet	QNN_DLC	float	Snapdragon® X Elite	0.958 ms	1 - 1 MB	NPU
GoogLeNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	0.559 ms	0 - 43 MB	NPU
GoogLeNet	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	1.752 ms	0 - 49 MB	NPU
GoogLeNet	QNN_DLC	float	Qualcomm® QCS8275	5.017 ms	1 - 26 MB	NPU
GoogLeNet	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	0.813 ms	1 - 78 MB	NPU
GoogLeNet	QNN_DLC	float	Qualcomm® QCS8450	1.752 ms	0 - 49 MB	NPU
GoogLeNet	QNN_DLC	float	Snapdragon® 8 Elite Mobile	0.454 ms	0 - 31 MB	NPU
GoogLeNet	QNN_DLC	float	Qualcomm® SA8295P	1.798 ms	0 - 24 MB	NPU
GoogLeNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.378 ms	1 - 32 MB	NPU
GoogLeNet	QNN_DLC	float	Qualcomm® SA7255P	5.017 ms	1 - 26 MB	NPU
GoogLeNet	QNN_DLC	float	Qualcomm® QCS9075	1.509 ms	3 - 5 MB	NPU
GoogLeNet	QNN_DLC	float	Qualcomm® QCS8750	0.454 ms	0 - 31 MB	NPU
GoogLeNet	QNN_DLC	float	Qualcomm® QCS7181	0.958 ms	1 - 1 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Snapdragon® X2 Elite	0.223 ms	0 - 0 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Snapdragon® X Elite	0.346 ms	0 - 0 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	0.191 ms	0 - 45 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Snapdragon® 8 Gen 1 Mobile	0.415 ms	0 - 50 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Qualcomm® QCS6490	1.033 ms	0 - 2 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Qualcomm® QCS8275	0.809 ms	0 - 29 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	0.244 ms	0 - 89 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Qualcomm® QCS8450	0.415 ms	0 - 50 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.138 ms	0 - 30 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	0.327 ms	0 - 31 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Qualcomm® QCM6690	2.363 ms	0 - 33 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Qualcomm® QCS9075	0.332 ms	0 - 2 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Qualcomm® SA7255P	0.809 ms	0 - 29 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Snapdragon® 8 Elite Mobile	0.153 ms	0 - 34 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Qualcomm® SA8295P	0.619 ms	0 - 28 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Qualcomm® QCS7790	0.327 ms	0 - 31 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Qualcomm® QCS8750	0.153 ms	0 - 34 MB	NPU
GoogLeNet	QNN_DLC	w8a8	Qualcomm® QCS7181	0.346 ms	0 - 0 MB	NPU
GoogLeNet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	0.561 ms	0 - 53 MB	NPU
GoogLeNet	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	1.764 ms	0 - 58 MB	NPU
GoogLeNet	TFLITE	float	Qualcomm® QCS8275	5.036 ms	0 - 33 MB	NPU
GoogLeNet	TFLITE	float	Qualcomm® QCS8550 (Proxy)	0.815 ms	0 - 2 MB	NPU
GoogLeNet	TFLITE	float	Qualcomm® SA8775P	10.851 ms	0 - 20 MB	GPU
GoogLeNet	TFLITE	float	Qualcomm® SA8650P	10.851 ms	0 - 20 MB	GPU
GoogLeNet	TFLITE	float	Qualcomm® SA8255P	10.851 ms	0 - 20 MB	GPU
GoogLeNet	TFLITE	float	Qualcomm® QCS8450	1.764 ms	0 - 58 MB	NPU
GoogLeNet	TFLITE	float	Snapdragon® 8 Elite Mobile	0.461 ms	0 - 38 MB	NPU
GoogLeNet	TFLITE	float	Qualcomm® SA8295P	1.802 ms	0 - 31 MB	NPU
GoogLeNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.373 ms	0 - 38 MB	NPU
GoogLeNet	TFLITE	float	Qualcomm® SA7255P	5.036 ms	0 - 33 MB	NPU
GoogLeNet	TFLITE	float	Qualcomm® QCS9075	1.529 ms	0 - 16 MB	NPU
GoogLeNet	TFLITE	float	Qualcomm® QCS8750	0.461 ms	0 - 38 MB	NPU
GoogLeNet	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	0.182 ms	0 - 42 MB	NPU
GoogLeNet	TFLITE	w8a8	Snapdragon® 8 Gen 1 Mobile	0.416 ms	0 - 48 MB	NPU
GoogLeNet	TFLITE	w8a8	Qualcomm® QCS6490	0.98 ms	0 - 8 MB	NPU
GoogLeNet	TFLITE	w8a8	Qualcomm® QCS8275	0.83 ms	0 - 28 MB	NPU
GoogLeNet	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	0.237 ms	0 - 18 MB	NPU
GoogLeNet	TFLITE	w8a8	Qualcomm® SA8775P	11.052 ms	1 - 23 MB	GPU
GoogLeNet	TFLITE	w8a8	Qualcomm® SA8650P	11.052 ms	1 - 23 MB	GPU
GoogLeNet	TFLITE	w8a8	Qualcomm® SA8255P	11.052 ms	1 - 23 MB	GPU
GoogLeNet	TFLITE	w8a8	Qualcomm® QCS8450	0.416 ms	0 - 48 MB	NPU
GoogLeNet	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.144 ms	0 - 29 MB	NPU
GoogLeNet	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	0.328 ms	0 - 29 MB	NPU
GoogLeNet	TFLITE	w8a8	Qualcomm® QCM6690	2.253 ms	0 - 29 MB	NPU
GoogLeNet	TFLITE	w8a8	Qualcomm® QCS9075	0.348 ms	0 - 9 MB	NPU
GoogLeNet	TFLITE	w8a8	Qualcomm® SA7255P	0.83 ms	0 - 28 MB	NPU
GoogLeNet	TFLITE	w8a8	Snapdragon® 8 Elite Mobile	0.162 ms	0 - 33 MB	NPU
GoogLeNet	TFLITE	w8a8	Qualcomm® SA8295P	0.654 ms	0 - 26 MB	NPU
GoogLeNet	TFLITE	w8a8	Qualcomm® QCS7790	0.328 ms	0 - 29 MB	NPU
GoogLeNet	TFLITE	w8a8	Qualcomm® QCS8750	0.162 ms	0 - 33 MB	NPU

License

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

References

• Going Deeper with Convolutions
• 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.