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 2.50, 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-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:

• Input resolution: 224x224
• Model checkpoint: Imagenet
• Model size (float): 339 MB
• Model size (w8a16): 90.2 MB
• Number of parameters: 88.8M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Swin-Base	ONNX	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	7.64 ms	0 - 398 MB	NPU
Swin-Base	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	9.496 ms	1 - 383 MB	NPU
Swin-Base	ONNX	float	Snapdragon® X2 Elite	7.826 ms	2 - 2 MB	NPU
Swin-Base	ONNX	float	Snapdragon® X Elite	19.463 ms	175 - 175 MB	NPU
Swin-Base	ONNX	float	Snapdragon® 8 Gen 3 Mobile	13.021 ms	0 - 538 MB	NPU
Swin-Base	ONNX	float	Snapdragon® 8 Gen 1 Mobile	28.691 ms	1 - 530 MB	NPU
Swin-Base	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	20.609 ms	0 - 5 MB	NPU
Swin-Base	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	18.857 ms	0 - 194 MB	NPU
Swin-Base	ONNX	float	Qualcomm® QCS8450	28.691 ms	1 - 530 MB	NPU
Swin-Base	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	21.825 ms	0 - 4 MB	NPU
Swin-Base	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	19.463 ms	175 - 175 MB	NPU
Swin-Base	ONNX	float	Qualcomm® Dragonwing™ Q-8750	9.496 ms	1 - 383 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	6.04 ms	0 - 414 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	8.057 ms	0 - 402 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® X2 Elite	6.213 ms	1 - 1 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® X Elite	16.078 ms	91 - 91 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	10.66 ms	0 - 542 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	20.508 ms	0 - 539 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	38.643 ms	0 - 4 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-8275	14.105 ms	0 - 5 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	15.525 ms	0 - 99 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® QCS8450	20.508 ms	0 - 539 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	16.281 ms	0 - 4 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	16.078 ms	91 - 91 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® Dragonwing™ Q-6690	86.579 ms	0 - 662 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	17.898 ms	0 - 606 MB	NPU
Swin-Base	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	8.057 ms	0 - 402 MB	NPU
Swin-Base	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	17.898 ms	0 - 606 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	7.201 ms	0 - 368 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	8.958 ms	0 - 352 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® X2 Elite	7.904 ms	1 - 1 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® X Elite	18.618 ms	1 - 1 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	12.209 ms	0 - 503 MB	NPU
Swin-Base	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	27.906 ms	0 - 493 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	20.082 ms	1 - 4 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	17.871 ms	1 - 3 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® QCS8450	27.906 ms	0 - 493 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	20.55 ms	1 - 3 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	18.618 ms	1 - 1 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	8.958 ms	0 - 352 MB	NPU
Swin-Base	QNN_DLC	float	Qualcomm® SA8295P	26.96 ms	1 - 343 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	7.132 ms	0 - 403 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	9.23 ms	0 - 387 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® X2 Elite	8.405 ms	0 - 0 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® X Elite	20.234 ms	0 - 0 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	12.993 ms	0 - 513 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	17.572 ms	0 - 4 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	19.313 ms	0 - 2 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® SA8650P	19.574 ms	0 - 411 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® SA8255P	19.574 ms	0 - 411 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	19.512 ms	0 - 3 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	20.234 ms	0 - 0 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	120.381 ms	0 - 640 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	21.734 ms	0 - 589 MB	NPU
Swin-Base	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	9.23 ms	0 - 387 MB	NPU
Swin-Base	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	21.734 ms	0 - 589 MB	NPU
Swin-Base	TFLITE	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	7.617 ms	0 - 388 MB	NPU
Swin-Base	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	9.299 ms	0 - 363 MB	NPU
Swin-Base	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	12.718 ms	0 - 507 MB	NPU
Swin-Base	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	28.204 ms	0 - 494 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	20.179 ms	0 - 179 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	18.647 ms	0 - 4 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® SA8775P	21.49 ms	0 - 359 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® SA8650P	21.49 ms	0 - 359 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® SA8255P	21.49 ms	0 - 359 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® QCS8450	28.204 ms	0 - 494 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	21.803 ms	0 - 178 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	9.299 ms	0 - 363 MB	NPU
Swin-Base	TFLITE	float	Qualcomm® SA8295P	27.902 ms	0 - 350 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.