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 1.24.1	Download
ONNX	w8a8	Universal	QAIRT 2.42, ONNX Runtime 1.24.1	Download
QNN_DLC	float	Universal	QAIRT 2.43	Download
QNN_DLC	w8a8	Universal	QAIRT 2.43	Download
TFLITE	float	Universal	QAIRT 2.43, TFLite 2.17.0	Download
TFLITE	w8a8	Universal	QAIRT 2.43, TFLite 2.17.0	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: 224x224
• 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	74.107 ms	6 - 329 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® X2 Elite	74.897 ms	53 - 53 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® X Elite	139.527 ms	53 - 53 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® 8 Gen 3 Mobile	109.976 ms	24 - 559 MB	NPU
Unet-Segmentation	ONNX	float	Qualcomm® QCS8550 (Proxy)	145.541 ms	16 - 19 MB	NPU
Unet-Segmentation	ONNX	float	Qualcomm® QCS9075	254.791 ms	9 - 22 MB	NPU
Unet-Segmentation	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	88.98 ms	14 - 330 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	16.609 ms	6 - 189 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® X2 Elite	20.132 ms	29 - 29 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® X Elite	39.081 ms	29 - 29 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	30.38 ms	6 - 340 MB	NPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS6490	4679.139 ms	942 - 1000 MB	CPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	41.85 ms	4 - 7 MB	NPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCS9075	35.607 ms	4 - 7 MB	NPU
Unet-Segmentation	ONNX	w8a8	Qualcomm® QCM6690	4136.243 ms	828 - 834 MB	CPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	24.71 ms	3 - 191 MB	NPU
Unet-Segmentation	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	3874.199 ms	841 - 847 MB	CPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	62.943 ms	0 - 343 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® X2 Elite	71.71 ms	9 - 9 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® X Elite	132.457 ms	9 - 9 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	102.107 ms	9 - 543 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	953.666 ms	1 - 323 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	133.686 ms	10 - 11 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA8775P	240.46 ms	1 - 324 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS9075	248.092 ms	9 - 27 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	274.935 ms	3 - 543 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA7255P	953.666 ms	1 - 323 MB	NPU
Unet-Segmentation	QNN_DLC	float	Qualcomm® SA8295P	274.504 ms	0 - 322 MB	NPU
Unet-Segmentation	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	82.531 ms	9 - 340 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	15.775 ms	6 - 201 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® X2 Elite	18.791 ms	2 - 2 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® X Elite	35.653 ms	2 - 2 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	26.174 ms	2 - 321 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS6490	267.73 ms	2 - 7 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	121.532 ms	2 - 181 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	34.815 ms	2 - 4 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA8775P	32.221 ms	1 - 180 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS9075	34.957 ms	4 - 10 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCM6690	1219.329 ms	2 - 520 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	60.072 ms	2 - 320 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA7255P	121.532 ms	2 - 181 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Qualcomm® SA8295P	63.759 ms	2 - 182 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	21.817 ms	2 - 189 MB	NPU
Unet-Segmentation	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	78.77 ms	2 - 271 MB	NPU
Unet-Segmentation	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	63.787 ms	6 - 352 MB	NPU
Unet-Segmentation	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	103.607 ms	6 - 543 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS8275 (Proxy)	954.39 ms	0 - 324 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS8550 (Proxy)	140.467 ms	0 - 305 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA8775P	240.498 ms	6 - 330 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS9075	248.114 ms	0 - 80 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® QCS8450 (Proxy)	274.79 ms	7 - 550 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA7255P	954.39 ms	0 - 324 MB	NPU
Unet-Segmentation	TFLITE	float	Qualcomm® SA8295P	274.548 ms	0 - 322 MB	NPU
Unet-Segmentation	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	83.251 ms	5 - 338 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	15.744 ms	6 - 202 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	26.316 ms	2 - 319 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS6490	267.809 ms	1 - 41 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	121.56 ms	2 - 181 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	32.332 ms	0 - 621 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA8775P	32.239 ms	2 - 180 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS9075	34.203 ms	0 - 36 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCM6690	1200.523 ms	1 - 519 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	58.435 ms	2 - 319 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA7255P	121.56 ms	2 - 181 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Qualcomm® SA8295P	63.784 ms	2 - 179 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	21.742 ms	2 - 186 MB	NPU
Unet-Segmentation	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	79.088 ms	2 - 270 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.