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	2.952 ms	0 - 242 MB	NPU
decoder	ONNX	float	Snapdragon® X2 Elite	3.024 ms	11 - 11 MB	NPU
decoder	ONNX	float	Snapdragon® X Elite	7.626 ms	11 - 11 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Gen 3 Mobile	4.959 ms	5 - 263 MB	NPU
decoder	ONNX	float	Qualcomm® QCS8550 (Proxy)	7.123 ms	0 - 15 MB	NPU
decoder	ONNX	float	Qualcomm® QCS9075	8.456 ms	4 - 7 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	3.668 ms	0 - 225 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	2.116 ms	3 - 191 MB	NPU
decoder	QNN_DLC	float	Snapdragon® X2 Elite	2.545 ms	4 - 4 MB	NPU
decoder	QNN_DLC	float	Snapdragon® X Elite	5.204 ms	4 - 4 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	3.361 ms	4 - 213 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	11.707 ms	0 - 183 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	4.833 ms	4 - 6 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8775P	5.854 ms	1 - 187 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS9075	5.891 ms	4 - 10 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	10.547 ms	0 - 212 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA7255P	11.707 ms	0 - 183 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8295P	7.323 ms	0 - 188 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	2.579 ms	0 - 187 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	2.605 ms	0 - 191 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	3.754 ms	0 - 243 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS8275 (Proxy)	12.829 ms	0 - 214 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS8550 (Proxy)	5.552 ms	0 - 8 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8775P	6.614 ms	0 - 216 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS9075	6.864 ms	0 - 19 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS8450 (Proxy)	9.278 ms	0 - 217 MB	NPU
decoder	TFLITE	float	Qualcomm® SA7255P	12.829 ms	0 - 214 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8295P	8.823 ms	0 - 191 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	2.99 ms	0 - 190 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	87.786 ms	20 - 1524 MB	NPU
encoder	ONNX	float	Snapdragon® X2 Elite	95.657 ms	14 - 14 MB	NPU
encoder	ONNX	float	Snapdragon® X Elite	171.215 ms	12 - 12 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Gen 3 Mobile	126.694 ms	16 - 3038 MB	NPU
encoder	ONNX	float	Qualcomm® QCS8550 (Proxy)	177.246 ms	12 - 42 MB	NPU
encoder	ONNX	float	Qualcomm® QCS9075	201.575 ms	12 - 15 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	101.243 ms	9 - 1396 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	45.026 ms	12 - 1312 MB	NPU
encoder	QNN_DLC	float	Snapdragon® X2 Elite	48.595 ms	12 - 12 MB	NPU
encoder	QNN_DLC	float	Snapdragon® X Elite	98.363 ms	12 - 12 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	66.345 ms	11 - 1559 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	209.768 ms	1 - 1193 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	95.439 ms	8 - 10 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8775P	101.195 ms	0 - 1601 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS9075	120.536 ms	12 - 30 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	316.624 ms	5 - 1574 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA7255P	209.768 ms	1 - 1193 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8295P	262.029 ms	0 - 1210 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	49.203 ms	11 - 1209 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	109.458 ms	32 - 1085 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	170.601 ms	23 - 2317 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS8275 (Proxy)	360.275 ms	33 - 1120 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS8550 (Proxy)	226.445 ms	33 - 37 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8775P	242.284 ms	33 - 1114 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS9075	294.179 ms	32 - 109 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS8450 (Proxy)	514.248 ms	0 - 1934 MB	NPU
encoder	TFLITE	float	Qualcomm® SA7255P	360.275 ms	33 - 1120 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8295P	373.937 ms	33 - 1160 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	130.308 ms	32 - 1113 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.