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 1.25.0	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	2 - 238 MB	NPU
decoder	ONNX	float	Snapdragon® X2 Elite	2.692 ms	208 - 208 MB	NPU
decoder	ONNX	float	Snapdragon® X Elite	6.197 ms	149 - 149 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Gen 3 Mobile	4.172 ms	0 - 260 MB	NPU
decoder	ONNX	float	Qualcomm® QCS8550 (Proxy)	5.929 ms	0 - 41 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	3.133 ms	0 - 234 MB	NPU
decoder	ONNX	float	Qualcomm® QCS9075	7.279 ms	4 - 49 MB	NPU
decoder	ONNX	float	Qualcomm® QCS8750	3.133 ms	0 - 234 MB	NPU
decoder	ONNX	float	Qualcomm® QCS7181	6.197 ms	149 - 149 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	2.528 ms	4 - 194 MB	NPU
decoder	QNN_DLC	float	Snapdragon® X2 Elite	3.033 ms	4 - 4 MB	NPU
decoder	QNN_DLC	float	Snapdragon® X Elite	6.674 ms	4 - 4 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	4.532 ms	4 - 222 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS8275	13.911 ms	2 - 194 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	6.312 ms	4 - 6 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8775P	7.26 ms	1 - 195 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8650P	7.26 ms	1 - 195 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8255P	7.26 ms	1 - 195 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	10.89 ms	0 - 225 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA7255P	13.911 ms	2 - 194 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8295P	9.245 ms	0 - 195 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	2.996 ms	0 - 190 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS9075	6.858 ms	4 - 10 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS8750	2.996 ms	0 - 190 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS7181	6.674 ms	4 - 4 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	2.142 ms	0 - 192 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	3.403 ms	0 - 212 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS8275	11.88 ms	0 - 186 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS8550 (Proxy)	4.967 ms	0 - 18 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8775P	5.944 ms	0 - 188 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8650P	5.944 ms	0 - 188 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8255P	5.944 ms	0 - 188 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS8450 (Proxy)	8.009 ms	0 - 218 MB	NPU
decoder	TFLITE	float	Qualcomm® SA7255P	11.88 ms	0 - 186 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8295P	7.272 ms	1 - 183 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	2.595 ms	0 - 184 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS9075	5.902 ms	0 - 19 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS8750	2.595 ms	0 - 184 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	87.726 ms	20 - 1015 MB	NPU
encoder	ONNX	float	Snapdragon® X2 Elite	93.916 ms	201 - 201 MB	NPU
encoder	ONNX	float	Snapdragon® X Elite	170.109 ms	169 - 169 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Gen 3 Mobile	125.651 ms	21 - 1193 MB	NPU
encoder	ONNX	float	Qualcomm® QCS8550 (Proxy)	175.807 ms	12 - 29 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	105.399 ms	8 - 1521 MB	NPU
encoder	ONNX	float	Qualcomm® QCS9075	199.5 ms	12 - 57 MB	NPU
encoder	ONNX	float	Qualcomm® QCS8750	105.399 ms	8 - 1521 MB	NPU
encoder	ONNX	float	Qualcomm® QCS7181	170.109 ms	169 - 169 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	41.597 ms	12 - 1244 MB	NPU
encoder	QNN_DLC	float	Snapdragon® X2 Elite	45.742 ms	12 - 12 MB	NPU
encoder	QNN_DLC	float	Snapdragon® X Elite	103.635 ms	12 - 12 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	66.235 ms	12 - 1331 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS8275	207.678 ms	2 - 1126 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	100.871 ms	12 - 481 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8775P	100.273 ms	1 - 1550 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8650P	100.273 ms	1 - 1550 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8255P	100.273 ms	1 - 1550 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	328.932 ms	12 - 1342 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA7255P	207.678 ms	2 - 1126 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8295P	325.055 ms	0 - 1556 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	48.691 ms	12 - 1127 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS9075	121.179 ms	12 - 30 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS8750	48.691 ms	12 - 1127 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS7181	103.635 ms	12 - 12 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	94.193 ms	32 - 1086 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	168.241 ms	24 - 2313 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS8275	364.551 ms	33 - 1119 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS8550 (Proxy)	221.19 ms	34 - 38 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8775P	241.191 ms	21 - 1102 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8650P	241.191 ms	21 - 1102 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8255P	241.191 ms	21 - 1102 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS8450 (Proxy)	520.217 ms	0 - 1935 MB	NPU
encoder	TFLITE	float	Qualcomm® SA7255P	364.551 ms	33 - 1119 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8295P	372.359 ms	34 - 1161 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	131.397 ms	33 - 1107 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS9075	293.282 ms	32 - 109 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS8750	131.397 ms	33 - 1107 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.