MobileSam: Optimized for Qualcomm Devices

Transformer based encoder-decoder where prompts specify what to segment in an image thereby allowing segmentation without the need for additional training. The image encoder generates embeddings and the lightweight decoder operates on the embeddings for point and mask based image segmentation.

This is based on the implementation of MobileSam 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 1.27.1	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit MobileSam 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 MobileSam on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

• Model checkpoint: vit_t
• Input resolution: 720p (720x1280)
• Number of parameters (sam_encoder): 6.95M
• Model size (sam_encoder) (float): 26.6 MB
• Number of parameters (sam_decoder): 6.16M
• Model size (sam_decoder) (float): 23.7 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
decoder	ONNX	float	Snapdragon® X2 Elite	2.636 ms	4 - 4 MB	NPU
decoder	ONNX	float	Snapdragon® X Elite	6.113 ms	11 - 11 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Gen 3 Mobile	4.15 ms	0 - 241 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Gen 1 Mobile	14.315 ms	5 - 233 MB	NPU
decoder	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	5.91 ms	4 - 6 MB	NPU
decoder	ONNX	float	Qualcomm® QCS8450	14.315 ms	5 - 233 MB	NPU
decoder	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	6.816 ms	4 - 7 MB	NPU
decoder	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	6.113 ms	11 - 11 MB	NPU
decoder	ONNX	float	Qualcomm® Dragonwing™ Q-8750	3.184 ms	0 - 237 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Elite Mobile	3.184 ms	0 - 237 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	2.544 ms	1 - 215 MB	NPU
decoder	QNN_DLC	float	Snapdragon® X2 Elite	3.005 ms	4 - 4 MB	NPU
decoder	QNN_DLC	float	Snapdragon® X Elite	5.711 ms	4 - 4 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	3.735 ms	4 - 223 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	11.599 ms	0 - 252 MB	NPU
decoder	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8275	12.588 ms	0 - 193 MB	NPU
decoder	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	5.279 ms	4 - 33 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8775P	6.269 ms	0 - 194 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8650P	6.269 ms	0 - 194 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8255P	6.269 ms	0 - 194 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS8450	11.599 ms	0 - 252 MB	NPU
decoder	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	6.05 ms	4 - 10 MB	NPU
decoder	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	5.711 ms	4 - 4 MB	NPU
decoder	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	2.835 ms	0 - 223 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA7255P	12.588 ms	0 - 193 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8295P	7.729 ms	0 - 191 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Elite Mobile	2.835 ms	0 - 223 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	2.537 ms	4 - 196 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	3.413 ms	0 - 211 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	8.008 ms	0 - 215 MB	NPU
decoder	TFLITE	float	Qualcomm® Dragonwing™ QCS8275	11.845 ms	0 - 185 MB	NPU
decoder	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	4.941 ms	0 - 17 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8775P	5.934 ms	0 - 189 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8650P	5.934 ms	0 - 189 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8255P	5.934 ms	0 - 189 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS8450	8.008 ms	0 - 215 MB	NPU
decoder	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	5.704 ms	0 - 19 MB	NPU
decoder	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	2.596 ms	0 - 184 MB	NPU
decoder	TFLITE	float	Qualcomm® SA7255P	11.845 ms	0 - 185 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8295P	7.282 ms	0 - 183 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Elite Mobile	2.596 ms	0 - 184 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	2.165 ms	0 - 187 MB	NPU
encoder	ONNX	float	Snapdragon® X2 Elite	100.973 ms	12 - 12 MB	NPU
encoder	ONNX	float	Snapdragon® X Elite	168.548 ms	14 - 14 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Gen 3 Mobile	125.733 ms	12 - 1103 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Gen 1 Mobile	407.934 ms	22 - 1128 MB	NPU
encoder	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	173.455 ms	12 - 30 MB	NPU
encoder	ONNX	float	Qualcomm® QCS8450	407.934 ms	22 - 1128 MB	NPU
encoder	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	178.544 ms	12 - 15 MB	NPU
encoder	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	168.548 ms	14 - 14 MB	NPU
encoder	ONNX	float	Qualcomm® Dragonwing™ Q-8750	100.448 ms	8 - 919 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Elite Mobile	100.448 ms	8 - 919 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	82.268 ms	9 - 1036 MB	NPU
encoder	QNN_DLC	float	Snapdragon® X2 Elite	46.051 ms	12 - 12 MB	NPU
encoder	QNN_DLC	float	Snapdragon® X Elite	103.021 ms	12 - 12 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	65.626 ms	12 - 1328 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	326.924 ms	11 - 1341 MB	NPU
encoder	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8275	207.442 ms	1 - 1126 MB	NPU
encoder	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	99.736 ms	12 - 15 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8775P	100.378 ms	1 - 1550 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8650P	100.378 ms	1 - 1550 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8255P	100.378 ms	1 - 1550 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS8450	326.924 ms	11 - 1341 MB	NPU
encoder	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	124.552 ms	14 - 32 MB	NPU
encoder	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	103.021 ms	12 - 12 MB	NPU
encoder	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	48.995 ms	6 - 1122 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA7255P	207.442 ms	1 - 1126 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8295P	324.927 ms	0 - 1557 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Elite Mobile	48.995 ms	6 - 1122 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	40.341 ms	12 - 1258 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	66.798 ms	0 - 3190 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	311.267 ms	4 - 1362 MB	NPU
encoder	TFLITE	float	Qualcomm® Dragonwing™ QCS8275	196.878 ms	4 - 1634 MB	NPU
encoder	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	90.526 ms	4 - 8 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8775P	100.683 ms	4 - 1670 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8650P	100.683 ms	4 - 1670 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8255P	100.683 ms	4 - 1670 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS8450	311.267 ms	4 - 1362 MB	NPU
encoder	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	101.504 ms	0 - 45 MB	NPU
encoder	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	48.492 ms	2 - 1629 MB	NPU
encoder	TFLITE	float	Qualcomm® SA7255P	196.878 ms	4 - 1634 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8295P	258.167 ms	4 - 1714 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Elite Mobile	48.492 ms	2 - 1629 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	40.214 ms	4 - 1796 MB	NPU

License

• The license for the original implementation of MobileSam can be found here.

References

• Segment Anything
• 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.