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 2.50, ONNX Runtime 1.30.0	Download
ONNX	w8a16	Universal	QAIRT 2.50, ONNX Runtime 1.30.0	Download
QNN_DLC	float	Universal	QAIRT 2.50	Download
QNN_DLC	w8a16	Universal	QAIRT 2.50	Download
TFLITE	float	Universal	QAIRT 2.50	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® 8 Elite Gen 5 For Galaxy Mobile	3.686 ms	0 - 187 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	4.414 ms	0 - 180 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® X2 Elite	3.845 ms	2 - 2 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® X Elite	9.475 ms	58 - 58 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Gen 3 Mobile	5.867 ms	0 - 252 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Gen 1 Mobile	14.097 ms	1 - 246 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	8.85 ms	0 - 5 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	8.993 ms	0 - 2 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® QCS8450	14.097 ms	1 - 246 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	9.988 ms	1 - 4 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	9.475 ms	58 - 58 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ Q-8750	4.414 ms	0 - 180 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	3.112 ms	0 - 214 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	3.97 ms	0 - 205 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® X2 Elite	3.266 ms	1 - 1 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® X Elite	8.364 ms	31 - 31 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	5.264 ms	0 - 285 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	9.62 ms	0 - 281 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	18.056 ms	0 - 4 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-8275	7.166 ms	0 - 4 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	8.002 ms	0 - 35 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® QCS8450	9.62 ms	0 - 281 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	8.338 ms	0 - 4 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	8.364 ms	31 - 31 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ Q-6690	34.805 ms	0 - 369 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	8.59 ms	0 - 229 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	3.97 ms	0 - 205 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	8.59 ms	0 - 229 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	3.563 ms	1 - 183 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	4.399 ms	1 - 175 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® X2 Elite	3.982 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® X Elite	9.636 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	5.821 ms	0 - 246 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	14.282 ms	0 - 240 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	8.72 ms	1 - 4 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	8.984 ms	1 - 2 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® QCS8450	14.282 ms	0 - 240 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	10.013 ms	3 - 5 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	9.636 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	4.399 ms	1 - 175 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA7255P	20.74 ms	1 - 172 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8295P	13.31 ms	1 - 172 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	3.648 ms	0 - 199 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	4.76 ms	0 - 191 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® X2 Elite	4.098 ms	0 - 0 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® X Elite	10.807 ms	0 - 0 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	6.569 ms	0 - 256 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	8.801 ms	0 - 3 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	9.98 ms	0 - 3 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA8650P	10.561 ms	0 - 205 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA8255P	10.561 ms	0 - 205 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	10.345 ms	2 - 5 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	10.807 ms	0 - 0 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	42.28 ms	0 - 362 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	10.571 ms	0 - 216 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	4.76 ms	0 - 191 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA7255P	17.79 ms	0 - 204 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	10.571 ms	0 - 216 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	3.856 ms	0 - 206 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	4.682 ms	0 - 185 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	6.076 ms	0 - 258 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	14.136 ms	0 - 248 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	8.978 ms	0 - 61 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	9.353 ms	0 - 3 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8775P	10.462 ms	0 - 189 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8650P	10.462 ms	0 - 189 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8255P	10.462 ms	0 - 189 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® QCS8450	14.136 ms	0 - 248 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	10.21 ms	0 - 61 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	4.682 ms	0 - 185 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA7255P	20.969 ms	0 - 188 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8295P	13.653 ms	0 - 183 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.