EfficientNet-B4: Optimized for Qualcomm Devices

EfficientNetB4 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 EfficientNet-B4 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 EfficientNet-B4 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 EfficientNet-B4 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_classification

Model Stats:

Model checkpoint: Imagenet
Input resolution: 380x380
Number of parameters: 19.3M
Model size (float): 73.6 MB
Model size (w8a16): 24.0 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
EfficientNet-B4	ONNX	float	Snapdragon® X2 Elite	3.918 ms	211 - 211 MB	NPU
EfficientNet-B4	ONNX	float	Snapdragon® X Elite	7.717 ms	177 - 177 MB	NPU
EfficientNet-B4	ONNX	float	Snapdragon® 8 Gen 3 Mobile	5.32 ms	2 - 139 MB	NPU
EfficientNet-B4	ONNX	float	Snapdragon® 8 Gen 1 Mobile	20.414 ms	2 - 179 MB	NPU
EfficientNet-B4	ONNX	float	Qualcomm® QCS8550 (Proxy)	7.34 ms	0 - 51 MB	NPU
EfficientNet-B4	ONNX	float	Qualcomm® QCS8450	20.414 ms	2 - 179 MB	NPU
EfficientNet-B4	ONNX	float	Snapdragon® 8 Elite Mobile	4.042 ms	0 - 89 MB	NPU
EfficientNet-B4	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	3.009 ms	2 - 203 MB	NPU
EfficientNet-B4	ONNX	float	Qualcomm® QCS9075	10.449 ms	2 - 47 MB	NPU
EfficientNet-B4	ONNX	float	Qualcomm® QCS8750	4.042 ms	0 - 89 MB	NPU
EfficientNet-B4	ONNX	float	Qualcomm® QCS7181	7.717 ms	177 - 177 MB	NPU
EfficientNet-B4	ONNX	w8a16	Snapdragon® X2 Elite	2.953 ms	212 - 212 MB	NPU
EfficientNet-B4	ONNX	w8a16	Snapdragon® X Elite	8.045 ms	148 - 148 MB	NPU
EfficientNet-B4	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	5.128 ms	1 - 213 MB	NPU
EfficientNet-B4	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	8.878 ms	1 - 218 MB	NPU
EfficientNet-B4	ONNX	w8a16	Qualcomm® QCS6490	26.085 ms	1 - 46 MB	NPU
EfficientNet-B4	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	7.661 ms	0 - 32 MB	NPU
EfficientNet-B4	ONNX	w8a16	Qualcomm® QCS8450	8.878 ms	1 - 218 MB	NPU
EfficientNet-B4	ONNX	w8a16	Qualcomm® QCS9075	8.335 ms	1 - 46 MB	NPU
EfficientNet-B4	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	2.771 ms	1 - 173 MB	NPU
EfficientNet-B4	ONNX	w8a16	Snapdragon® 8 Elite Mobile	3.47 ms	0 - 167 MB	NPU
EfficientNet-B4	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	8.941 ms	1 - 286 MB	NPU
EfficientNet-B4	ONNX	w8a16	Qualcomm® QCM6690	53.977 ms	1 - 294 MB	NPU
EfficientNet-B4	ONNX	w8a16	Qualcomm® QCS7790	8.941 ms	1 - 286 MB	NPU
EfficientNet-B4	ONNX	w8a16	Qualcomm® QCS8750	3.47 ms	0 - 167 MB	NPU
EfficientNet-B4	ONNX	w8a16	Qualcomm® QCS7181	8.045 ms	148 - 148 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® X2 Elite	4.597 ms	2 - 2 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® X Elite	8.817 ms	2 - 2 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	5.823 ms	0 - 141 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	23.067 ms	1 - 185 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® QCS8275	29.129 ms	2 - 81 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	8.157 ms	2 - 166 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® QCS8450	23.067 ms	1 - 185 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® 8 Elite Mobile	4.314 ms	0 - 87 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® SA8295P	18.8 ms	2 - 124 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	3.258 ms	2 - 207 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® SA7255P	29.129 ms	2 - 81 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® QCS9075	11.701 ms	2 - 5 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® QCS8750	4.314 ms	0 - 87 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® QCS7181	8.817 ms	2 - 2 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® X2 Elite	3.544 ms	1 - 1 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® X Elite	9.093 ms	1 - 1 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	5.67 ms	1 - 199 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® 8 Gen 1 Mobile	11.258 ms	1 - 200 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS6490	23.01 ms	1 - 3 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS8275	15.484 ms	1 - 141 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	8.397 ms	1 - 183 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS8450	11.258 ms	1 - 200 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS9075	9.88 ms	2 - 5 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	3.04 ms	1 - 157 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	3.749 ms	0 - 144 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® SA8295P	10.959 ms	1 - 144 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	9.765 ms	1 - 268 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCM6690	56.876 ms	1 - 276 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® SA7255P	15.484 ms	1 - 141 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS7790	9.765 ms	1 - 268 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS8750	3.749 ms	0 - 144 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS7181	9.093 ms	1 - 1 MB	NPU
EfficientNet-B4	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	5.756 ms	0 - 163 MB	NPU
EfficientNet-B4	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	21.772 ms	0 - 204 MB	NPU
EfficientNet-B4	TFLITE	float	Qualcomm® QCS8275	28.921 ms	0 - 97 MB	NPU
EfficientNet-B4	TFLITE	float	Qualcomm® QCS8550 (Proxy)	8.024 ms	0 - 4 MB	NPU
EfficientNet-B4	TFLITE	float	Qualcomm® SA8775P	29.409 ms	0 - 32 MB	GPU
EfficientNet-B4	TFLITE	float	Qualcomm® SA8650P	29.409 ms	0 - 32 MB	GPU
EfficientNet-B4	TFLITE	float	Qualcomm® SA8255P	29.409 ms	0 - 32 MB	GPU
EfficientNet-B4	TFLITE	float	Qualcomm® QCS8450	21.772 ms	0 - 204 MB	NPU
EfficientNet-B4	TFLITE	float	Snapdragon® 8 Elite Mobile	4.322 ms	0 - 104 MB	NPU
EfficientNet-B4	TFLITE	float	Qualcomm® SA8295P	18.833 ms	0 - 139 MB	NPU
EfficientNet-B4	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	3.218 ms	0 - 98 MB	NPU
EfficientNet-B4	TFLITE	float	Qualcomm® SA7255P	28.921 ms	0 - 97 MB	NPU
EfficientNet-B4	TFLITE	float	Qualcomm® QCS9075	10.939 ms	0 - 49 MB	NPU
EfficientNet-B4	TFLITE	float	Qualcomm® QCS8750	4.322 ms	0 - 104 MB	NPU

License

The license for the original implementation of EfficientNet-B4 can be found here.

References

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.

Downloads last month: -; Downloads are not tracked for this model. How to track

Paper for qualcomm/EfficientNet-B4

EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks

Paper • 1905.11946 • Published May 28, 2019 • 4