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 2.42, ONNX Runtime 1.25.0	Download
ONNX	w8a16	Universal	QAIRT 2.42, 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-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:

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

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 Mobile	3.657 ms	0 - 455 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® X2 Elite	3.928 ms	212 - 212 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® X Elite	9.226 ms	181 - 181 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Gen 3 Mobile	5.784 ms	1 - 605 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® QCS8550 (Proxy)	8.931 ms	0 - 81 MB	NPU
Swin-Tiny	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	4.455 ms	0 - 445 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® QCS9075	11.152 ms	1 - 46 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® QCS8750	4.455 ms	0 - 445 MB	NPU
Swin-Tiny	ONNX	float	Qualcomm® QCS7181	9.226 ms	181 - 181 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	3.44 ms	0 - 234 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® X2 Elite	3.698 ms	181 - 181 MB	NPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	5.472 ms	0 - 334 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® QCS6490	414.595 ms	95 - 107 MB	CPU
Swin-Tiny	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	8.334 ms	0 - 43 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® QCM6690	207.913 ms	102 - 118 MB	CPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	193.187 ms	103 - 119 MB	CPU
Swin-Tiny	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	4.263 ms	0 - 266 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® QCS9075	9.891 ms	0 - 45 MB	NPU
Swin-Tiny	ONNX	w8a16	Qualcomm® QCS7790	193.187 ms	103 - 119 MB	CPU
Swin-Tiny	ONNX	w8a16	Qualcomm® QCS8750	4.263 ms	0 - 266 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	3.745 ms	1 - 410 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® X2 Elite	4.282 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® X Elite	9.86 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	6.015 ms	0 - 245 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® QCS8275	21.27 ms	1 - 176 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	9.211 ms	1 - 3 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8775P	10.458 ms	1 - 437 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8650P	10.458 ms	1 - 437 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8255P	10.458 ms	1 - 437 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	14.497 ms	0 - 241 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA7255P	21.27 ms	1 - 176 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® SA8295P	13.887 ms	1 - 433 MB	NPU
Swin-Tiny	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	4.55 ms	0 - 414 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® QCS9075	11.408 ms	1 - 3 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® QCS8750	4.55 ms	0 - 414 MB	NPU
Swin-Tiny	QNN_DLC	float	Qualcomm® QCS7181	9.86 ms	1 - 1 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	3.814 ms	0 - 434 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® X2 Elite	4.371 ms	0 - 0 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® X Elite	10.829 ms	0 - 0 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	6.576 ms	0 - 509 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® QCS8275	17.71 ms	0 - 434 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	10.162 ms	0 - 193 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA8775P	10.61 ms	0 - 442 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA8650P	10.61 ms	0 - 442 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA8255P	10.61 ms	0 - 442 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® QCM6690	42.39 ms	0 - 359 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® SA7255P	17.71 ms	0 - 434 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	10.636 ms	0 - 438 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	4.834 ms	0 - 425 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® QCS9075	11.827 ms	0 - 2 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® QCS7790	10.636 ms	0 - 438 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® QCS8750	4.834 ms	0 - 425 MB	NPU
Swin-Tiny	QNN_DLC	w8a16	Qualcomm® QCS7181	10.829 ms	0 - 0 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	3.807 ms	0 - 210 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	6.043 ms	0 - 268 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® QCS8275	21.246 ms	0 - 200 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® QCS8550 (Proxy)	9.24 ms	0 - 4 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8775P	10.443 ms	0 - 200 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8650P	10.443 ms	0 - 200 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8255P	10.443 ms	0 - 200 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® QCS8450 (Proxy)	14.367 ms	0 - 258 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA7255P	21.246 ms	0 - 200 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® SA8295P	13.907 ms	0 - 198 MB	NPU
Swin-Tiny	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	4.625 ms	0 - 195 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® QCS9075	11.685 ms	0 - 60 MB	NPU
Swin-Tiny	TFLITE	float	Qualcomm® QCS8750	4.625 ms	0 - 195 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.