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 1.25.0	Download
ONNX	w8a8	Universal	QAIRT 2.42, ONNX Runtime 1.25.0	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	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	70.679 ms	22 - 348 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® X2 Elite	75.065 ms	202 - 202 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® X Elite	139.85 ms	139 - 139 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® 8 Gen 3 Mobile	109.283 ms	25 - 563 MB	NPU
Unet-Segmentation	ONNX	float	Qualcomm® QCS8550 (Proxy)	146.541 ms	21 - 944 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	88.419 ms	15 - 332 MB	NPU
Unet-Segmentation	ONNX	float	Qualcomm® QCS9075	254.684 ms	9 - 64 MB	NPU
Unet-Segmentation	ONNX	float	Qualcomm® QCS8750	88.419 ms	15 - 332 MB	NPU
Unet-Segmentation	ONNX	float	Qualcomm® QCS7181	139.85 ms	139 - 139 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	16.53 ms	5 - 192 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® X2 Elite	20.137 ms	210 - 210 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® X Elite	38.97 ms	179 - 179 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	30.135 ms	6 - 342 MB	NPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS6490	4659.834 ms	943 - 1000 MB	CPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	39.25 ms	0 - 46 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	3891.089 ms	843 - 850 MB	CPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCM6690	4131.885 ms	834 - 840 MB	CPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	24.427 ms	3 - 190 MB	NPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS9075	35.562 ms	4 - 49 MB	NPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS7790	3891.089 ms	843 - 850 MB	CPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS8750	24.427 ms	3 - 190 MB	NPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS7181	38.97 ms	179 - 179 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	62.723 ms	9 - 353 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® X2 Elite	71.838 ms	9 - 9 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® X Elite	132.956 ms	9 - 9 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	102.113 ms	8 - 523 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS8275	953.369 ms	0 - 324 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	133.833 ms	10 - 819 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA8775P	240.391 ms	1 - 325 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA8650P	240.391 ms	1 - 325 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA8255P	240.391 ms	1 - 325 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	275.728 ms	3 - 540 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA8295P	274.455 ms	0 - 322 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	83.509 ms	0 - 332 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA7255P	953.369 ms	0 - 324 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS9075	248.045 ms	9 - 27 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS8750	83.509 ms	0 - 332 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS7181	132.956 ms	9 - 9 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	15.68 ms	4 - 200 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® X2 Elite	18.9 ms	2 - 2 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® X Elite	35.738 ms	2 - 2 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	26.103 ms	2 - 317 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS6490	267.604 ms	2 - 8 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS8275	121.521 ms	1 - 180 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	34.892 ms	2 - 4 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA8775P	32.174 ms	2 - 182 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA8650P	32.174 ms	2 - 182 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA8255P	32.174 ms	2 - 182 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA7255P	121.521 ms	1 - 180 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	81.381 ms	3 - 273 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA8295P	63.756 ms	0 - 180 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCM6690	1214.163 ms	3 - 525 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	21.938 ms	2 - 188 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS9075	35.063 ms	1 - 6 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	58.614 ms	2 - 318 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS7790	81.381 ms	3 - 273 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS8750	21.938 ms	2 - 188 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS7181	35.738 ms	2 - 2 MB	NPU
Unet-Segmentation	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	66.931 ms	6 - 350 MB	NPU
Unet-Segmentation	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	103.252 ms	5 - 576 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS8275	953.515 ms	4 - 328 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS8550 (Proxy)	138.086 ms	7 - 219 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA8775P	240.462 ms	7 - 330 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA8650P	240.462 ms	7 - 330 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA8255P	240.462 ms	7 - 330 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS8450 (Proxy)	276.393 ms	5 - 585 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA8295P	274.42 ms	7 - 328 MB	NPU
Unet-Segmentation	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	82.458 ms	6 - 335 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA7255P	953.515 ms	4 - 328 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS9075	248.046 ms	0 - 80 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS8750	82.458 ms	6 - 335 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	15.79 ms	0 - 196 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	26.317 ms	1 - 318 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS6490	267.779 ms	0 - 40 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS8275	121.535 ms	2 - 181 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	34.844 ms	2 - 4 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA8775P	32.225 ms	2 - 180 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA8650P	32.225 ms	2 - 180 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA8255P	32.225 ms	2 - 180 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA7255P	121.535 ms	2 - 181 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	78.758 ms	2 - 264 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA8295P	63.767 ms	2 - 180 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCM6690	1237.067 ms	0 - 519 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	21.89 ms	0 - 187 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS9075	34.236 ms	0 - 36 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	57.758 ms	2 - 317 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS7790	78.758 ms	2 - 264 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS8750	21.89 ms	0 - 187 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.