SINet: Optimized for Qualcomm Devices

SINet is a machine learning model that is designed to segment people from close-up portrait images in real time.

This is based on the implementation of SINet 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
ONNX	w8a8	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
QNN_DLC	w8a8	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download
TFLITE	w8a8	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit SINet 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 SINet on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

Model checkpoint: SINet.pth
Input resolution: 224x224
Number of output classes: 2 (foreground / background)
Number of parameters: 91.9K
Model size (float): 415 KB
Model size (w8a8): 241 KB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
SINet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.707 ms	0 - 41 MB	NPU
SINet	ONNX	float	Snapdragon® X2 Elite	0.772 ms	212 - 212 MB	NPU
SINet	ONNX	float	Snapdragon® X Elite	1.739 ms	149 - 149 MB	NPU
SINet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	1.107 ms	0 - 48 MB	NPU
SINet	ONNX	float	Qualcomm® QCS8550 (Proxy)	1.695 ms	0 - 12 MB	NPU
SINet	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	0.859 ms	0 - 34 MB	NPU
SINet	ONNX	float	Qualcomm® QCS9075	2.138 ms	2 - 47 MB	NPU
SINet	ONNX	float	Qualcomm® QCS8750	0.859 ms	0 - 34 MB	NPU
SINet	ONNX	float	Qualcomm® QCS7181	1.739 ms	149 - 149 MB	NPU
SINet	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	0.866 ms	0 - 48 MB	NPU
SINet	ONNX	w8a16	Snapdragon® X2 Elite	0.94 ms	212 - 212 MB	NPU
SINet	ONNX	w8a16	Snapdragon® X Elite	2.009 ms	149 - 149 MB	NPU
SINet	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	1.27 ms	0 - 58 MB	NPU
SINet	ONNX	w8a16	Qualcomm® QCS6490	53.101 ms	14 - 17 MB	CPU
SINet	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	1.938 ms	0 - 48 MB	NPU
SINet	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	1.017 ms	0 - 43 MB	NPU
SINet	ONNX	w8a16	Qualcomm® QCM6690	19.951 ms	21 - 29 MB	CPU
SINet	ONNX	w8a16	Qualcomm® QCS9075	2.212 ms	0 - 46 MB	NPU
SINet	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	13.597 ms	21 - 30 MB	CPU
SINet	ONNX	w8a16	Qualcomm® QCS7790	13.597 ms	21 - 30 MB	CPU
SINet	ONNX	w8a16	Qualcomm® QCS8750	1.017 ms	0 - 43 MB	NPU
SINet	ONNX	w8a16	Qualcomm® QCS7181	2.009 ms	149 - 149 MB	NPU
SINet	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.494 ms	0 - 46 MB	NPU
SINet	ONNX	w8a8	Snapdragon® X2 Elite	0.502 ms	212 - 212 MB	NPU
SINet	ONNX	w8a8	Snapdragon® X Elite	1.135 ms	181 - 181 MB	NPU
SINet	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	0.749 ms	0 - 53 MB	NPU
SINet	ONNX	w8a8	Qualcomm® QCS6490	38.214 ms	7 - 11 MB	CPU
SINet	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	1.091 ms	0 - 14 MB	NPU
SINet	ONNX	w8a8	Qualcomm® QCS9075	1.361 ms	0 - 45 MB	NPU
SINet	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	11.307 ms	7 - 16 MB	CPU
SINet	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	0.586 ms	0 - 46 MB	NPU
SINet	ONNX	w8a8	Qualcomm® QCM6690	16.036 ms	7 - 16 MB	CPU
SINet	ONNX	w8a8	Qualcomm® QCS7790	11.307 ms	7 - 16 MB	CPU
SINet	ONNX	w8a8	Qualcomm® QCS8750	0.586 ms	0 - 46 MB	NPU
SINet	ONNX	w8a8	Qualcomm® QCS7181	1.135 ms	181 - 181 MB	NPU
SINet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	1.124 ms	1 - 38 MB	NPU
SINet	QNN_DLC	float	Snapdragon® X2 Elite	1.425 ms	1 - 1 MB	NPU
SINet	QNN_DLC	float	Snapdragon® X Elite	3.806 ms	1 - 1 MB	NPU
SINet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	2.101 ms	0 - 46 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS8275	6.636 ms	1 - 34 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	3.433 ms	1 - 2 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA8775P	3.91 ms	1 - 36 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA8650P	3.91 ms	1 - 36 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA8255P	3.91 ms	1 - 36 MB	NPU
SINet	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	1.805 ms	1 - 33 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	3.746 ms	0 - 52 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA7255P	6.636 ms	1 - 34 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS9075	3.833 ms	1 - 4 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA8295P	4.658 ms	1 - 36 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS8750	1.805 ms	1 - 33 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS7181	3.806 ms	1 - 1 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	3.204 ms	0 - 42 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® X2 Elite	3.528 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® X Elite	7.131 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	4.068 ms	0 - 49 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS8275	10.792 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	6.569 ms	0 - 2 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA8775P	7.061 ms	0 - 41 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA8650P	7.061 ms	0 - 41 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA8255P	7.061 ms	0 - 41 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	3.437 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA7255P	10.792 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA8295P	8.139 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS9075	6.847 ms	0 - 3 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS8450 (Proxy)	7.184 ms	0 - 51 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS8750	3.437 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS7181	7.131 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	2.14 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® X2 Elite	2.347 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® X Elite	4.708 ms	1 - 1 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	2.905 ms	24 - 71 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS8275	7.75 ms	0 - 36 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	4.494 ms	0 - 12 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA8775P	4.681 ms	0 - 39 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA8650P	4.681 ms	0 - 39 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA8255P	4.681 ms	0 - 39 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS9075	4.453 ms	0 - 2 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	4.636 ms	0 - 49 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	2.363 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA8295P	5.226 ms	0 - 35 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA7255P	7.75 ms	0 - 36 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS8750	2.363 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS7181	4.708 ms	1 - 1 MB	NPU
SINet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	1.128 ms	0 - 39 MB	NPU
SINet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	2.116 ms	0 - 45 MB	NPU
SINet	TFLITE	float	Qualcomm® QCS8275	6.636 ms	1 - 34 MB	NPU
SINet	TFLITE	float	Qualcomm® QCS8550 (Proxy)	3.393 ms	0 - 2 MB	NPU
SINet	TFLITE	float	Qualcomm® SA8775P	3.902 ms	0 - 37 MB	NPU
SINet	TFLITE	float	Qualcomm® SA8650P	3.902 ms	0 - 37 MB	NPU
SINet	TFLITE	float	Qualcomm® SA8255P	3.902 ms	0 - 37 MB	NPU
SINet	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	1.807 ms	0 - 30 MB	NPU
SINet	TFLITE	float	Qualcomm® QCS8450 (Proxy)	3.782 ms	0 - 49 MB	NPU
SINet	TFLITE	float	Qualcomm® SA7255P	6.636 ms	1 - 34 MB	NPU
SINet	TFLITE	float	Qualcomm® QCS9075	3.843 ms	1 - 4 MB	NPU
SINet	TFLITE	float	Qualcomm® SA8295P	4.67 ms	1 - 34 MB	NPU
SINet	TFLITE	float	Qualcomm® QCS8750	1.807 ms	0 - 30 MB	NPU

License

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

References

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.

Downloads last month: -; Downloads are not tracked for this model. How to track

Paper for qualcomm/SINet

SINet: Extreme Lightweight Portrait Segmentation Networks with Spatial Squeeze Modules and Information Blocking Decoder

Paper • 1911.09099 • Published Nov 20, 2019