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 1.27.1	Download
ONNX	w8a16	Universal	QAIRT 2.50, ONNX Runtime 1.27.1	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	3.521 ms	0 - 195 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	4.327 ms	0 - 190 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® X2 Elite	3.62 ms	2 - 2 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® X Elite	9.061 ms	57 - 57 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Gen 3 Mobile	5.729 ms	0 - 260 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Gen 1 Mobile	13.689 ms	0 - 253 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	8.632 ms	0 - 5 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	8.7 ms	0 - 66 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® QCS8450	13.689 ms	0 - 253 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	9.591 ms	0 - 4 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	9.061 ms	57 - 57 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® Dragonwing™ Q-8750	4.327 ms	0 - 190 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	2.967 ms	0 - 215 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	3.851 ms	0 - 205 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® X2 Elite	3.123 ms	1 - 1 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® X Elite	8.134 ms	31 - 31 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	5.209 ms	0 - 288 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	9.536 ms	0 - 280 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	16.892 ms	0 - 4 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-8275	6.941 ms	0 - 4 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	7.741 ms	0 - 35 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® QCS8450	9.536 ms	0 - 280 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	8.111 ms	0 - 4 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	8.134 ms	31 - 31 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ Q-6690	33.533 ms	0 - 375 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	8.371 ms	0 - 230 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	3.851 ms	0 - 205 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	8.371 ms	0 - 230 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	3.564 ms	0 - 182 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	4.385 ms	0 - 172 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® X2 Elite	4.044 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® X Elite	9.681 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	5.836 ms	0 - 247 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	14.293 ms	0 - 238 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)	8.921 ms	1 - 4 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8775P	10.224 ms	1 - 174 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8650P	10.224 ms	1 - 174 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8255P	10.224 ms	1 - 174 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® QCS8450	14.293 ms	0 - 238 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	10.007 ms	1 - 3 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	9.681 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	4.385 ms	0 - 172 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA7255P	20.747 ms	1 - 172 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8295P	13.295 ms	1 - 172 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	3.638 ms	0 - 200 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	4.731 ms	0 - 196 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® X2 Elite	4.107 ms	0 - 0 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® X Elite	10.828 ms	0 - 0 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	6.516 ms	0 - 259 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	8.799 ms	0 - 3 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	10.01 ms	0 - 2 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA8775P	10.561 ms	0 - 205 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	10.352 ms	0 - 3 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	10.828 ms	0 - 0 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	38.802 ms	0 - 362 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	10.69 ms	0 - 217 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	4.731 ms	0 - 196 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	10.69 ms	0 - 217 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	3.688 ms	0 - 206 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	4.496 ms	0 - 192 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	5.91 ms	0 - 260 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	14.011 ms	0 - 247 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	8.851 ms	0 - 61 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	9.068 ms	0 - 3 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8775P	10.236 ms	0 - 189 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8650P	10.236 ms	0 - 189 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8255P	10.236 ms	0 - 189 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® QCS8450	14.011 ms	0 - 247 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	9.983 ms	0 - 61 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	4.496 ms	0 - 192 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA7255P	85.924 ms	1 - 28 MB	GPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8295P	13.487 ms	0 - 184 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.