Swin-Base: Optimized for Qualcomm Devices

SwinBase 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 Swin-Base 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 Swin-Base 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 Swin-Base 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: 88.8M
• Model size (float): 339 MB
• Model size (w8a16): 90.2 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Swin-Base	ONNX	float	Snapdragon® X2 Elite	8.416 ms	180 - 180 MB	NPU
Swin-Base	ONNX	float	Snapdragon® X Elite	19.615 ms	203 - 203 MB	NPU
Swin-Base	ONNX	float	Snapdragon® 8 Gen 3 Mobile	12.887 ms	1 - 528 MB	NPU
Swin-Base	ONNX	float	Snapdragon® 8 Gen 1 Mobile	29.576 ms	1 - 515 MB	NPU
Swin-Base	ONNX	float	Qualcomm® QCS8550 (Proxy)	18.844 ms	0 - 196 MB	NPU
Swin-Base	ONNX	float	Qualcomm® QCS8450	29.576 ms	1 - 515 MB	NPU
Swin-Base	ONNX	float	Snapdragon® 8 Elite Mobile	9.77 ms	1 - 388 MB	NPU
Swin-Base	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	7.938 ms	1 - 407 MB	NPU
Swin-Base	ONNX	float	Qualcomm® QCS9075	24.016 ms	1 - 46 MB	NPU
Swin-Base	ONNX	float	Qualcomm® QCS8750	9.77 ms	1 - 388 MB	NPU
Swin-Base	ONNX	float	Qualcomm® QCS7181	19.615 ms	203 - 203 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® X2 Elite	6.96 ms	212 - 212 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® X Elite	16.211 ms	149 - 149 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	10.398 ms	0 - 531 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	15.369 ms	0 - 100 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® QCS9075	19.07 ms	0 - 45 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	6.498 ms	0 - 431 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® 8 Elite Mobile	8.153 ms	0 - 418 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	17.628 ms	0 - 617 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® QCM6690	86.041 ms	0 - 688 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® QCS7790	17.628 ms	0 - 617 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® QCS8750	8.153 ms	0 - 418 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® QCS7181	16.211 ms	149 - 149 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® X2 Elite	8.521 ms	1 - 1 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® X Elite	19.292 ms	1 - 1 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	12.512 ms	0 - 516 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	28.95 ms	1 - 504 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® QCS8275	53.618 ms	1 - 363 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	18.322 ms	1 - 3 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® QCS8450	28.95 ms	1 - 504 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® 8 Elite Mobile	9.434 ms	1 - 365 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® SA8295P	27.617 ms	1 - 353 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	7.512 ms	0 - 386 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® SA7255P	53.618 ms	1 - 363 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® QCS9075	23.01 ms	3 - 5 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® QCS8750	9.434 ms	1 - 365 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® QCS7181	19.292 ms	1 - 1 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® X2 Elite	8.504 ms	0 - 0 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® X Elite	20.471 ms	0 - 0 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	12.877 ms	0 - 507 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® QCS8275	35.382 ms	0 - 411 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	19.179 ms	0 - 347 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® QCS9075	22.541 ms	0 - 2 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	7.607 ms	0 - 423 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	9.542 ms	0 - 406 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	21.538 ms	0 - 606 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® QCM6690	123.093 ms	0 - 930 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® SA7255P	35.382 ms	0 - 411 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® QCS7790	21.538 ms	0 - 606 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® QCS8750	9.542 ms	0 - 406 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® QCS7181	20.471 ms	0 - 0 MB	NPU
Swin-Base	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	12.836 ms	0 - 1055 MB	NPU
Swin-Base	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	28.937 ms	0 - 511 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® QCS8275	53.538 ms	0 - 706 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® QCS8550 (Proxy)	18.947 ms	0 - 4 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® SA8775P	111.156 ms	0 - 34 MB	GPU
Swin-Base	TFLITE	float	Qualcomm® SA8650P	111.156 ms	0 - 34 MB	GPU
Swin-Base	TFLITE	float	Qualcomm® SA8255P	111.156 ms	0 - 34 MB	GPU
Swin-Base	TFLITE	float	Qualcomm® QCS8450	28.937 ms	0 - 511 MB	NPU
Swin-Base	TFLITE	float	Snapdragon® 8 Elite Mobile	9.746 ms	0 - 387 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® SA8295P	28.374 ms	0 - 372 MB	NPU
Swin-Base	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	7.923 ms	0 - 398 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® SA7255P	53.538 ms	0 - 706 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® QCS9075	24.108 ms	0 - 178 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® QCS8750	9.746 ms	0 - 387 MB	NPU

License

• The license for the original implementation of Swin-Base can be found here.

References

• Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
• 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.