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 2.42, ONNX Runtime 1.24.3	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® 8 Elite Gen 5 Mobile	1.467 ms	1 - 78 MB	NPU
EfficientNet-B4	ONNX	float	Snapdragon® 8 Elite Mobile	1.759 ms	0 - 71 MB	NPU
EfficientNet-B4	ONNX	float	Snapdragon® X2 Elite	1.628 ms	45 - 45 MB	NPU
EfficientNet-B4	ONNX	float	Snapdragon® X Elite	3.346 ms	45 - 45 MB	NPU
EfficientNet-B4	ONNX	float	Snapdragon® X Elite	3.346 ms	45 - 45 MB	NPU
EfficientNet-B4	ONNX	float	Snapdragon® 8 Gen 3 Mobile	2.271 ms	0 - 129 MB	NPU
EfficientNet-B4	ONNX	float	Qualcomm® QCS8550 (Proxy)	3.055 ms	0 - 51 MB	NPU
EfficientNet-B4	ONNX	float	Qualcomm® QCS9075	4.023 ms	0 - 4 MB	NPU
EfficientNet-B4	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	1.759 ms	0 - 71 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	1.507 ms	0 - 68 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® 8 Elite Mobile	1.842 ms	0 - 69 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® X2 Elite	1.941 ms	1 - 1 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® X Elite	3.599 ms	1 - 1 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® X Elite	3.599 ms	1 - 1 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	2.385 ms	0 - 117 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	12.006 ms	1 - 65 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	3.347 ms	0 - 30 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® QCS9075	4.132 ms	1 - 3 MB	NPU
EfficientNet-B4	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	7.865 ms	0 - 136 MB	NPU
EfficientNet-B4	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	1.842 ms	0 - 69 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	1.317 ms	0 - 109 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	1.595 ms	0 - 104 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® X2 Elite	1.701 ms	0 - 0 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® X Elite	3.763 ms	0 - 0 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® X Elite	3.763 ms	0 - 0 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	2.292 ms	0 - 147 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS6490	8.757 ms	2 - 4 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS8275 (Proxy)	6.565 ms	0 - 101 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	3.447 ms	0 - 2 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS9075	3.78 ms	0 - 2 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCM6690	16.121 ms	0 - 232 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Qualcomm® QCS8450 (Proxy)	4.191 ms	0 - 151 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	1.595 ms	0 - 104 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	3.565 ms	0 - 107 MB	NPU
EfficientNet-B4	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	3.565 ms	0 - 107 MB	NPU
EfficientNet-B4	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	1.509 ms	0 - 85 MB	NPU
EfficientNet-B4	TFLITE	float	Snapdragon® 8 Elite Mobile	1.842 ms	0 - 87 MB	NPU
EfficientNet-B4	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	2.397 ms	0 - 145 MB	NPU
EfficientNet-B4	TFLITE	float	Qualcomm® QCS8275 (Proxy)	12.043 ms	0 - 82 MB	NPU
EfficientNet-B4	TFLITE	float	Qualcomm® QCS8550 (Proxy)	3.307 ms	0 - 2 MB	NPU
EfficientNet-B4	TFLITE	float	Qualcomm® QCS9075	4.157 ms	0 - 48 MB	NPU
EfficientNet-B4	TFLITE	float	Qualcomm® QCS8450 (Proxy)	7.877 ms	0 - 162 MB	NPU
EfficientNet-B4	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	1.842 ms	0 - 87 MB	NPU

License

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

References

• EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
• 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.