Swin-Tiny: Optimized for Qualcomm Devices

SwinTiny 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-Tiny 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 1.27.1	Download
ONNX	w8a16	Universal	QAIRT 2.45, ONNX Runtime 1.27.1	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-Tiny 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-Tiny on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_classification

Model Stats:

• Input resolution: 224x224
• Model checkpoint: Imagenet
• Model size (float): 110 MB
• Model size (w8a16): 29.9 MB
• Number of parameters: 28.8M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Swin-Tiny	ONNX	float	Snapdragon® X2 Elite	4.052 ms	2 - 2 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® X Elite	9.476 ms	58 - 58 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Gen 3 Mobile	5.92 ms	1 - 269 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Gen 1 Mobile	13.83 ms	0 - 255 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	8.914 ms	0 - 4 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	8.887 ms	0 - 85 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® QCS8450	13.83 ms	0 - 255 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	9.906 ms	0 - 4 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	9.476 ms	58 - 58 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ Q-8750	4.511 ms	0 - 443 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Elite Mobile	4.511 ms	0 - 443 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	3.691 ms	0 - 449 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® X2 Elite	3.466 ms	1 - 1 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® X Elite	8.222 ms	31 - 31 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	5.135 ms	0 - 330 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	9.702 ms	0 - 276 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	24.544 ms	0 - 3 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-8275	7.081 ms	0 - 4 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	7.8 ms	0 - 248 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® QCS8450	9.702 ms	0 - 276 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	8.144 ms	0 - 3 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	8.222 ms	31 - 31 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ Q-6690	34.677 ms	0 - 385 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	8.259 ms	0 - 235 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	3.981 ms	0 - 265 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Elite Mobile	3.981 ms	0 - 265 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	3.268 ms	0 - 273 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	8.259 ms	0 - 235 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® X2 Elite	4.468 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® X Elite	9.943 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	5.976 ms	0 - 246 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	14.524 ms	0 - 241 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	9.003 ms	1 - 4 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	9.129 ms	1 - 208 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8775P	10.433 ms	1 - 435 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8650P	10.433 ms	1 - 435 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8255P	10.433 ms	1 - 435 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® QCS8450	14.524 ms	0 - 241 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	10.265 ms	3 - 5 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	9.943 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	4.521 ms	0 - 414 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA7255P	21.27 ms	1 - 175 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8295P	13.782 ms	0 - 433 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Elite Mobile	4.521 ms	0 - 414 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	3.766 ms	1 - 414 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® X2 Elite	4.414 ms	0 - 0 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® X Elite	10.928 ms	0 - 0 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	6.63 ms	0 - 510 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	8.827 ms	0 - 3 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	10.131 ms	0 - 195 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA8775P	10.637 ms	0 - 442 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA8650P	10.637 ms	0 - 442 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA8255P	10.637 ms	0 - 442 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	10.501 ms	0 - 2 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	10.928 ms	0 - 0 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	42.762 ms	0 - 425 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	10.673 ms	0 - 435 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	4.895 ms	0 - 431 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA7255P	17.8 ms	0 - 437 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	4.895 ms	0 - 431 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	3.892 ms	0 - 437 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	10.673 ms	0 - 435 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	6.067 ms	0 - 272 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	14.336 ms	0 - 254 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	9.055 ms	0 - 61 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	9.253 ms	0 - 5 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8775P	10.438 ms	0 - 201 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8650P	10.438 ms	0 - 201 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8255P	10.438 ms	0 - 201 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® QCS8450	14.336 ms	0 - 254 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	10.178 ms	0 - 60 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	4.604 ms	0 - 199 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA7255P	21.213 ms	0 - 199 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8295P	13.895 ms	0 - 199 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Elite Mobile	4.604 ms	0 - 199 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	3.836 ms	0 - 213 MB	NPU

License

• The license for the original implementation of Swin-Tiny 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.