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 2.42, ONNX Runtime 1.24.3	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® 8 Elite Gen 5 Mobile	2.622 ms	0 - 244 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Elite Mobile	3.134 ms	0 - 234 MB	NPU
decoder	ONNX	float	Snapdragon® X2 Elite	2.686 ms	11 - 11 MB	NPU
decoder	ONNX	float	Snapdragon® X Elite	6.359 ms	11 - 11 MB	NPU
decoder	ONNX	float	Snapdragon® X Elite	6.359 ms	11 - 11 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Gen 3 Mobile	4.146 ms	0 - 263 MB	NPU
decoder	ONNX	float	Qualcomm® QCS8550 (Proxy)	5.972 ms	0 - 15 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	3.134 ms	0 - 234 MB	NPU
decoder	ONNX	float	Qualcomm® QCS9075	7.345 ms	4 - 7 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	2.535 ms	3 - 192 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Elite Mobile	2.856 ms	0 - 190 MB	NPU
decoder	QNN_DLC	float	Snapdragon® X2 Elite	2.991 ms	4 - 4 MB	NPU
decoder	QNN_DLC	float	Snapdragon® X Elite	5.708 ms	4 - 4 MB	NPU
decoder	QNN_DLC	float	Snapdragon® X Elite	5.708 ms	4 - 4 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	3.704 ms	4 - 225 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	5.327 ms	4 - 6 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8775P	6.228 ms	1 - 192 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8775P	6.228 ms	1 - 192 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8775P	6.228 ms	1 - 192 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA7255P	12.586 ms	1 - 193 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8295P	7.7 ms	0 - 195 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	2.856 ms	0 - 190 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS9075	6.363 ms	4 - 10 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	11.541 ms	0 - 224 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	2.145 ms	0 - 192 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Elite Mobile	2.592 ms	0 - 189 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	3.41 ms	0 - 210 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS8550 (Proxy)	4.967 ms	0 - 8 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8775P	5.93 ms	0 - 187 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8775P	5.93 ms	0 - 187 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8775P	5.93 ms	0 - 187 MB	NPU
decoder	TFLITE	float	Qualcomm® SA7255P	11.877 ms	0 - 185 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8295P	7.262 ms	0 - 183 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	2.592 ms	0 - 189 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS9075	5.954 ms	0 - 19 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS8450 (Proxy)	8.045 ms	0 - 217 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	80.351 ms	20 - 1519 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Elite Mobile	96.875 ms	8 - 1394 MB	NPU
encoder	ONNX	float	Snapdragon® X2 Elite	101.697 ms	14 - 14 MB	NPU
encoder	ONNX	float	Snapdragon® X Elite	167.909 ms	12 - 12 MB	NPU
encoder	ONNX	float	Snapdragon® X Elite	167.909 ms	12 - 12 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Gen 3 Mobile	120.888 ms	20 - 3034 MB	NPU
encoder	ONNX	float	Qualcomm® QCS8550 (Proxy)	171.47 ms	12 - 32 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	96.875 ms	8 - 1394 MB	NPU
encoder	ONNX	float	Qualcomm® QCS9075	198.958 ms	12 - 15 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	41.397 ms	12 - 1244 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Elite Mobile	48.646 ms	10 - 1124 MB	NPU
encoder	QNN_DLC	float	Snapdragon® X2 Elite	45.794 ms	12 - 12 MB	NPU
encoder	QNN_DLC	float	Snapdragon® X Elite	102.913 ms	12 - 12 MB	NPU
encoder	QNN_DLC	float	Snapdragon® X Elite	102.913 ms	12 - 12 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	65.629 ms	0 - 1317 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	99.402 ms	10 - 13 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8775P	100.321 ms	0 - 1549 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8775P	100.321 ms	0 - 1549 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8775P	100.321 ms	0 - 1549 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA7255P	207.539 ms	1 - 1127 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8295P	325.068 ms	0 - 1556 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	48.646 ms	10 - 1124 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS9075	120.717 ms	14 - 32 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	329.911 ms	12 - 1342 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	109.604 ms	32 - 1086 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Elite Mobile	129.101 ms	25 - 1105 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	166.461 ms	33 - 2313 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS8550 (Proxy)	218.803 ms	38 - 42 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8775P	242.262 ms	34 - 1115 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8775P	242.262 ms	34 - 1115 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8775P	242.262 ms	34 - 1115 MB	NPU
encoder	TFLITE	float	Qualcomm® SA7255P	377.475 ms	33 - 1120 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8295P	372.998 ms	22 - 1145 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	129.101 ms	25 - 1105 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS9075	299.389 ms	32 - 109 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS8450 (Proxy)	537.18 ms	0 - 1926 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.