Unet-Segmentation: Optimized for Qualcomm Devices

UNet is a machine learning model that produces a segmentation mask for an image. The most basic use case will label each pixel in the image as being in the foreground or the background. More advanced usage will assign a class label to each pixel. This version of the model was trained on the data from Kaggle's Carvana Image Masking Challenge (see https://www.kaggle.com/c/carvana-image-masking-challenge) and is used for vehicle segmentation.

This is based on the implementation of Unet-Segmentation 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 1.24.3	Download
ONNX	w8a8	Universal	QAIRT 2.42, ONNX Runtime 1.24.3	Download
QNN_DLC	float	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 Unet-Segmentation 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 Unet-Segmentation on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

• Model checkpoint: unet_carvana_scale1.0_epoch2
• Input resolution: 640x1280
• Number of output classes: 2 (foreground / background)
• Number of parameters: 31.0M
• Model size (float): 118 MB
• Model size (w8a8): 29.8 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Unet-Segmentation	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	69.648 ms	5 - 328 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® 8 Elite Mobile	90.468 ms	15 - 333 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® X2 Elite	75.285 ms	53 - 53 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® X Elite	139.427 ms	53 - 53 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® X Elite	139.427 ms	53 - 53 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® 8 Gen 3 Mobile	110.568 ms	2 - 537 MB	NPU
Unet-Segmentation	ONNX	float	Qualcomm® QCS8550 (Proxy)	160.361 ms	0 - 452 MB	NPU
Unet-Segmentation	ONNX	float	Qualcomm® QCS9075	254.769 ms	9 - 21 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	90.468 ms	15 - 333 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	16.43 ms	0 - 184 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 8 Elite Mobile	24.865 ms	3 - 191 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® X2 Elite	20.214 ms	29 - 29 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® X Elite	39.064 ms	29 - 29 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® X Elite	39.064 ms	29 - 29 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	30.177 ms	6 - 337 MB	NPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS6490	4673.96 ms	945 - 1002 MB	CPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	40.195 ms	4 - 6 MB	NPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS9075	35.651 ms	4 - 7 MB	NPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCM6690	4187.869 ms	828 - 835 MB	CPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	24.865 ms	3 - 191 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	3889.111 ms	842 - 850 MB	CPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	3889.111 ms	842 - 850 MB	CPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	61.809 ms	9 - 352 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® 8 Elite Mobile	82.364 ms	9 - 342 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® X2 Elite	71.862 ms	9 - 9 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® X Elite	132.376 ms	9 - 9 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® X Elite	132.376 ms	9 - 9 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	103.692 ms	8 - 522 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	954.479 ms	0 - 323 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	142.471 ms	9 - 857 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA8775P	240.426 ms	0 - 323 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA8775P	240.426 ms	0 - 323 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA8775P	240.426 ms	0 - 323 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS9075	248.031 ms	9 - 27 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	291.567 ms	4 - 541 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA7255P	954.479 ms	0 - 323 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA8295P	274.454 ms	0 - 322 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	82.364 ms	9 - 342 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	15.753 ms	2 - 198 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 8 Elite Mobile	21.891 ms	2 - 188 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® X2 Elite	18.848 ms	2 - 2 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® X Elite	35.651 ms	2 - 2 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® X Elite	35.651 ms	2 - 2 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	26.104 ms	2 - 319 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS6490	267.641 ms	2 - 8 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	121.54 ms	1 - 180 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	35.022 ms	2 - 5 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA8775P	32.194 ms	1 - 180 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA8775P	32.194 ms	1 - 180 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA8775P	32.194 ms	1 - 180 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS9075	34.633 ms	1 - 6 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCM6690	1202.116 ms	2 - 522 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	58.446 ms	2 - 318 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA7255P	121.54 ms	1 - 180 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA8295P	63.715 ms	2 - 181 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	21.891 ms	2 - 188 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	78.739 ms	2 - 268 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	78.739 ms	2 - 268 MB	NPU
Unet-Segmentation	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	66.075 ms	6 - 345 MB	NPU
Unet-Segmentation	TFLITE	float	Snapdragon® 8 Elite Mobile	82.297 ms	5 - 336 MB	NPU
Unet-Segmentation	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	105.484 ms	5 - 575 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS8275 (Proxy)	953.329 ms	0 - 323 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS8550 (Proxy)	146.708 ms	6 - 107 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA8775P	240.513 ms	6 - 330 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA8775P	240.513 ms	6 - 330 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA8775P	240.513 ms	6 - 330 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS9075	248.057 ms	0 - 80 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS8450 (Proxy)	317.123 ms	8 - 588 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA7255P	953.329 ms	0 - 323 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA8295P	274.502 ms	6 - 328 MB	NPU
Unet-Segmentation	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	82.297 ms	5 - 336 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	15.882 ms	2 - 197 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 8 Elite Mobile	21.907 ms	1 - 188 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	26.167 ms	1 - 315 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS6490	267.608 ms	2 - 41 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	121.525 ms	2 - 180 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	35.161 ms	2 - 399 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA8775P	32.215 ms	2 - 180 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA8775P	32.215 ms	2 - 180 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA8775P	32.215 ms	2 - 180 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS9075	34.207 ms	0 - 36 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCM6690	1241.466 ms	0 - 521 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	60.085 ms	2 - 316 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA7255P	121.525 ms	2 - 180 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA8295P	63.734 ms	2 - 181 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	21.907 ms	1 - 188 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	78.528 ms	0 - 264 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	78.528 ms	0 - 264 MB	NPU

License

• The license for the original implementation of Unet-Segmentation can be found here.

References

• U-Net: Convolutional Networks for Biomedical Image Segmentation
• 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.