RepViT: Optimized for Qualcomm Devices

RepViT is a lightweight pure CNN model for mobile devices, incorporating efficient architectural designs from Vision Transformers into CNNs. It achieves over 80% top-1 accuracy on ImageNet with 1ms latency on an iPhone 12.

This is based on the implementation of RepViT 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
QNN_DLC	float	Universal	QAIRT 2.50	Download
TFLITE	float	Universal	QAIRT 2.50	Download

For more device-specific assets and performance metrics, visit RepViT 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 RepViT 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): 91 MB
• Number of parameters: 22.9M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
RepViT	ONNX	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	1.43 ms	0 - 114 MB	NPU
RepViT	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	1.776 ms	0 - 113 MB	NPU
RepViT	ONNX	float	Snapdragon® X2 Elite	1.697 ms	2 - 2 MB	NPU
RepViT	ONNX	float	Snapdragon® X Elite	3.327 ms	48 - 48 MB	NPU
RepViT	ONNX	float	Snapdragon® 8 Gen 3 Mobile	2.356 ms	0 - 194 MB	NPU
RepViT	ONNX	float	Snapdragon® 8 Gen 1 Mobile	11.235 ms	0 - 192 MB	NPU
RepViT	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	4.608 ms	0 - 5 MB	NPU
RepViT	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	3.192 ms	0 - 50 MB	NPU
RepViT	ONNX	float	Qualcomm® QCS8450	11.235 ms	0 - 192 MB	NPU
RepViT	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	4.334 ms	0 - 4 MB	NPU
RepViT	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	3.327 ms	48 - 48 MB	NPU
RepViT	ONNX	float	Qualcomm® Dragonwing™ Q-8750	1.776 ms	0 - 113 MB	NPU
RepViT	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	1.519 ms	1 - 106 MB	NPU
RepViT	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	1.887 ms	0 - 101 MB	NPU
RepViT	QNN_DLC	float	Snapdragon® X2 Elite	2.057 ms	1 - 1 MB	NPU
RepViT	QNN_DLC	float	Snapdragon® X Elite	3.864 ms	1 - 1 MB	NPU
RepViT	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	2.529 ms	1 - 182 MB	NPU
RepViT	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	11.884 ms	1 - 175 MB	NPU
RepViT	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	4.73 ms	1 - 4 MB	NPU
RepViT	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	3.476 ms	1 - 3 MB	NPU
RepViT	QNN_DLC	float	Qualcomm® SA8775P	4.728 ms	1 - 103 MB	NPU
RepViT	QNN_DLC	float	Qualcomm® SA8650P	4.728 ms	1 - 103 MB	NPU
RepViT	QNN_DLC	float	Qualcomm® SA8255P	4.728 ms	1 - 103 MB	NPU
RepViT	QNN_DLC	float	Qualcomm® QCS8450	11.884 ms	1 - 175 MB	NPU
RepViT	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	4.556 ms	3 - 5 MB	NPU
RepViT	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	3.864 ms	1 - 1 MB	NPU
RepViT	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	1.887 ms	0 - 101 MB	NPU
RepViT	QNN_DLC	float	Qualcomm® SA8295P	9.508 ms	1 - 103 MB	NPU
RepViT	TFLITE	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	1.507 ms	0 - 169 MB	NPU
RepViT	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	1.876 ms	0 - 161 MB	NPU
RepViT	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	2.51 ms	0 - 230 MB	NPU
RepViT	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	11.937 ms	0 - 224 MB	NPU
RepViT	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	4.746 ms	0 - 51 MB	NPU
RepViT	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	3.429 ms	0 - 3 MB	NPU
RepViT	TFLITE	float	Qualcomm® SA8775P	4.696 ms	0 - 164 MB	NPU
RepViT	TFLITE	float	Qualcomm® SA8650P	4.696 ms	0 - 164 MB	NPU
RepViT	TFLITE	float	Qualcomm® SA8255P	4.696 ms	0 - 164 MB	NPU
RepViT	TFLITE	float	Qualcomm® QCS8450	11.937 ms	0 - 224 MB	NPU
RepViT	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	4.571 ms	0 - 51 MB	NPU
RepViT	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	1.876 ms	0 - 161 MB	NPU
RepViT	TFLITE	float	Qualcomm® SA8295P	9.497 ms	0 - 159 MB	NPU

License

• The license for the original implementation of RepViT can be found here.

References

• RepViT: Revisiting Mobile CNN From ViT Perspective
• 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.