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 2.45, 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

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® X2 Elite	0.763 ms	2 - 2 MB	NPU
SINet	ONNX	float	Snapdragon® X Elite	1.765 ms	2 - 2 MB	NPU
SINet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	1.133 ms	0 - 47 MB	NPU
SINet	ONNX	float	Snapdragon® 8 Gen 1 Mobile	2.027 ms	1 - 56 MB	NPU
SINet	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	1.695 ms	0 - 9 MB	NPU
SINet	ONNX	float	Qualcomm® QCS8450	2.027 ms	1 - 56 MB	NPU
SINet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.707 ms	0 - 37 MB	NPU
SINet	ONNX	float	Snapdragon® 8 Elite Mobile	0.84 ms	0 - 33 MB	NPU
SINet	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	2.572 ms	1 - 5 MB	NPU
SINet	ONNX	float	Qualcomm® Dragonwing™ Q-8750	0.84 ms	0 - 33 MB	NPU
SINet	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	1.765 ms	2 - 2 MB	NPU
SINet	QNN_DLC	float	Snapdragon® X2 Elite	1.4 ms	1 - 1 MB	NPU
SINet	QNN_DLC	float	Snapdragon® X Elite	3.81 ms	1 - 1 MB	NPU
SINet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	2.103 ms	0 - 46 MB	NPU
SINet	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	3.735 ms	0 - 55 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS8275	6.609 ms	1 - 35 MB	NPU
SINet	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	3.435 ms	1 - 3 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA8775P	3.916 ms	0 - 36 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA8650P	3.916 ms	0 - 36 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA8255P	3.916 ms	0 - 36 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS8450	3.735 ms	0 - 55 MB	NPU
SINet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	1.129 ms	1 - 36 MB	NPU
SINet	QNN_DLC	float	Snapdragon® 8 Elite Mobile	1.805 ms	0 - 32 MB	NPU
SINet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	3.709 ms	1 - 4 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA7255P	6.609 ms	1 - 35 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA8295P	4.665 ms	1 - 36 MB	NPU
SINet	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	1.805 ms	0 - 32 MB	NPU
SINet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	3.81 ms	1 - 1 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® X2 Elite	3.527 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® X Elite	7.178 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	4.063 ms	0 - 49 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® 8 Gen 1 Mobile	7.134 ms	0 - 55 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS8275	10.804 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.586 ms	0 - 2 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA8775P	7.07 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA8650P	7.07 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA8255P	7.07 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS8450	7.134 ms	0 - 55 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	3.436 ms	0 - 41 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	6.795 ms	0 - 3 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA7255P	10.804 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA8295P	8.124 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	3.21 ms	0 - 39 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	3.436 ms	0 - 41 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	7.178 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® X2 Elite	2.343 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® X Elite	4.729 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	2.909 ms	0 - 48 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® 8 Gen 1 Mobile	4.626 ms	0 - 53 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS8275	7.73 ms	0 - 35 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® Dragonwing™ QCS8550 (Proxy)	4.482 ms	0 - 2 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA8775P	4.693 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA8650P	4.693 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA8255P	4.693 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS8450	4.626 ms	0 - 53 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® 8 Elite Mobile	2.36 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® Dragonwing™ IQ-9075	4.829 ms	2 - 4 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA7255P	7.73 ms	0 - 35 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA8295P	5.228 ms	0 - 35 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	2.136 ms	0 - 38 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® Dragonwing™ Q-8750	2.36 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® Dragonwing™ IQ-X7181	4.729 ms	0 - 0 MB	NPU
SINet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	2.109 ms	0 - 44 MB	NPU
SINet	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	3.77 ms	0 - 54 MB	NPU
SINet	TFLITE	float	Qualcomm® QCS8275	6.673 ms	1 - 35 MB	NPU
SINet	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	3.397 ms	0 - 17 MB	NPU
SINet	TFLITE	float	Qualcomm® SA8775P	3.881 ms	0 - 37 MB	NPU
SINet	TFLITE	float	Qualcomm® SA8650P	3.881 ms	0 - 37 MB	NPU
SINet	TFLITE	float	Qualcomm® SA8255P	3.881 ms	0 - 37 MB	NPU
SINet	TFLITE	float	Qualcomm® QCS8450	3.77 ms	0 - 54 MB	NPU
SINet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	1.136 ms	0 - 34 MB	NPU
SINet	TFLITE	float	Snapdragon® 8 Elite Mobile	1.775 ms	0 - 30 MB	NPU
SINet	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	3.706 ms	1 - 4 MB	NPU
SINet	TFLITE	float	Qualcomm® SA7255P	6.673 ms	1 - 35 MB	NPU
SINet	TFLITE	float	Qualcomm® SA8295P	4.649 ms	1 - 35 MB	NPU
SINet	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	1.775 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