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	0.761 ms	180 - 180 MB	NPU
SINet	ONNX	float	Snapdragon® X Elite	1.768 ms	148 - 148 MB	NPU
SINet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	1.121 ms	1 - 37 MB	NPU
SINet	ONNX	float	Snapdragon® 8 Gen 1 Mobile	2.031 ms	1 - 49 MB	NPU
SINet	ONNX	float	Qualcomm® QCS8550 (Proxy)	1.681 ms	0 - 80 MB	NPU
SINet	ONNX	float	Qualcomm® QCS8450	2.031 ms	1 - 49 MB	NPU
SINet	ONNX	float	Snapdragon® 8 Elite Mobile	0.838 ms	0 - 34 MB	NPU
SINet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.708 ms	0 - 34 MB	NPU
SINet	ONNX	float	Qualcomm® QCS9075	2.133 ms	2 - 47 MB	NPU
SINet	ONNX	float	Qualcomm® QCS8750	0.838 ms	0 - 34 MB	NPU
SINet	ONNX	float	Qualcomm® QCS7181	1.768 ms	148 - 148 MB	NPU
SINet	QNN_DLC	float	Snapdragon® X2 Elite	1.386 ms	1 - 1 MB	NPU
SINet	QNN_DLC	float	Snapdragon® X Elite	3.843 ms	1 - 1 MB	NPU
SINet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	2.108 ms	0 - 46 MB	NPU
SINet	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	3.743 ms	0 - 50 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS8275	6.656 ms	1 - 34 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	3.428 ms	1 - 2 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS8450	3.743 ms	0 - 50 MB	NPU
SINet	QNN_DLC	float	Snapdragon® 8 Elite Mobile	1.809 ms	1 - 38 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA7255P	6.656 ms	1 - 34 MB	NPU
SINet	QNN_DLC	float	Qualcomm® SA8295P	4.637 ms	0 - 35 MB	NPU
SINet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	1.127 ms	1 - 36 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS9075	3.832 ms	1 - 4 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS8750	1.809 ms	1 - 38 MB	NPU
SINet	QNN_DLC	float	Qualcomm® QCS7181	3.843 ms	1 - 1 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® X2 Elite	3.509 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® X Elite	7.13 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	4.069 ms	0 - 49 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® 8 Gen 1 Mobile	7.151 ms	0 - 58 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS8275	10.793 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	6.553 ms	0 - 3 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS8450	7.151 ms	0 - 58 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	3.215 ms	0 - 39 MB	NPU
SINet	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	3.438 ms	0 - 41 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS9075	6.859 ms	0 - 3 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA8295P	8.126 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® SA7255P	10.793 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS8750	3.438 ms	0 - 41 MB	NPU
SINet	QNN_DLC	w8a16	Qualcomm® QCS7181	7.13 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® X2 Elite	2.337 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® X Elite	4.731 ms	0 - 0 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	2.895 ms	0 - 47 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® 8 Gen 1 Mobile	4.62 ms	0 - 51 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS8275	7.754 ms	0 - 36 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	4.488 ms	0 - 2 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS8450	4.62 ms	0 - 51 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS9075	4.446 ms	0 - 2 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA7255P	7.754 ms	0 - 36 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® SA8295P	5.223 ms	0 - 35 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	2.139 ms	0 - 37 MB	NPU
SINet	QNN_DLC	w8a8	Snapdragon® 8 Elite Mobile	2.362 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS8750	2.362 ms	0 - 40 MB	NPU
SINet	QNN_DLC	w8a8	Qualcomm® QCS7181	4.731 ms	0 - 0 MB	NPU
SINet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	2.108 ms	0 - 45 MB	NPU
SINet	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	3.779 ms	0 - 53 MB	NPU
SINet	TFLITE	float	Qualcomm® QCS8275	6.646 ms	1 - 36 MB	NPU
SINet	TFLITE	float	Qualcomm® QCS8550 (Proxy)	3.401 ms	0 - 2 MB	NPU
SINet	TFLITE	float	Qualcomm® SA8775P	5.413 ms	2 - 18 MB	CPU
SINet	TFLITE	float	Qualcomm® SA8650P	5.413 ms	2 - 18 MB	CPU
SINet	TFLITE	float	Qualcomm® SA8255P	5.413 ms	2 - 18 MB	CPU
SINet	TFLITE	float	Qualcomm® QCS8450	3.779 ms	0 - 53 MB	NPU
SINet	TFLITE	float	Snapdragon® 8 Elite Mobile	1.807 ms	0 - 36 MB	NPU
SINet	TFLITE	float	Qualcomm® SA7255P	6.646 ms	1 - 36 MB	NPU
SINet	TFLITE	float	Qualcomm® SA8295P	4.678 ms	0 - 34 MB	NPU
SINet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	1.137 ms	0 - 35 MB	NPU
SINet	TFLITE	float	Qualcomm® QCS9075	3.843 ms	1 - 4 MB	NPU
SINet	TFLITE	float	Qualcomm® QCS8750	1.807 ms	0 - 36 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