Segment-Anything-Model-3: Optimized for Qualcomm Devices

SAM3 (Segment Anything with Concepts) extends SAM2 with open-vocabulary segmentation, producing bounding boxes and masks for objects matching a natural-language prompt.

This is based on the implementation of Segment-Anything-Model-3 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

Due to licensing restrictions, we cannot distribute pre-exported model assets for this model. 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

See our repository for Segment-Anything-Model-3 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

• Input resolution: 1008x1008
• Model checkpoint: sam3
• Model size (SAM3Backbone) (float): 128 MB
• Model size (SAM3Transformer) (float): 23.7 MB
• Number of parameters (SAM3Backbone): 33.5M
• Number of parameters (SAM3Transformer): 6.22M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
head	PRECOMPILED_QNN_ONNX	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	199.555 ms	131 - 144 MB	NPU
head	PRECOMPILED_QNN_ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	294.323 ms	130 - 142 MB	NPU
head	PRECOMPILED_QNN_ONNX	float	Snapdragon® X2 Elite	192.599 ms	136 - 136 MB	NPU
head	PRECOMPILED_QNN_ONNX	float	Snapdragon® X Elite	490.286 ms	664 - 664 MB	NPU
head	PRECOMPILED_QNN_ONNX	float	Snapdragon® 8 Gen 3 Mobile	403.684 ms	131 - 140 MB	NPU
head	PRECOMPILED_QNN_ONNX	float	Qualcomm® Dragonwing™ IQ-8275	496.148 ms	106 - 217 MB	NPU
head	PRECOMPILED_QNN_ONNX	float	Qualcomm® Dragonwing™ IQ-9075	503.347 ms	106 - 216 MB	NPU
head	PRECOMPILED_QNN_ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	490.286 ms	664 - 664 MB	NPU
head	PRECOMPILED_QNN_ONNX	float	Qualcomm® Dragonwing™ Q-8750	294.323 ms	130 - 142 MB	NPU
head	QNN_CONTEXT_BINARY	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	215.164 ms	82 - 90 MB	NPU
head	QNN_CONTEXT_BINARY	float	Snapdragon® 8 Elite For Galaxy Mobile	294.579 ms	81 - 93 MB	NPU
head	QNN_CONTEXT_BINARY	float	Snapdragon® X2 Elite	191.35 ms	107 - 107 MB	NPU
head	QNN_CONTEXT_BINARY	float	Snapdragon® X Elite	496.047 ms	107 - 107 MB	NPU
head	QNN_CONTEXT_BINARY	float	Snapdragon® 8 Gen 3 Mobile	391.799 ms	106 - 119 MB	NPU
head	QNN_CONTEXT_BINARY	float	Qualcomm® SA8650P	516.497 ms	97 - 107 MB	NPU
head	QNN_CONTEXT_BINARY	float	Qualcomm® SA8255P	516.497 ms	97 - 107 MB	NPU
head	QNN_CONTEXT_BINARY	float	Qualcomm® Dragonwing™ IQ-9075	502.313 ms	106 - 278 MB	NPU
head	QNN_CONTEXT_BINARY	float	Qualcomm® Dragonwing™ IQ-X7181	496.047 ms	107 - 107 MB	NPU
head	QNN_CONTEXT_BINARY	float	Qualcomm® Dragonwing™ Q-8750	294.579 ms	81 - 93 MB	NPU
head	QNN_CONTEXT_BINARY	float	Qualcomm® SA8295P	545.962 ms	106 - 112 MB	NPU
vision_backbone	PRECOMPILED_QNN_ONNX	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	1097.891 ms	9 - 21 MB	NPU
vision_backbone	PRECOMPILED_QNN_ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	1295.951 ms	8 - 19 MB	NPU
vision_backbone	PRECOMPILED_QNN_ONNX	float	Snapdragon® X2 Elite	1006.596 ms	265 - 265 MB	NPU
vision_backbone	PRECOMPILED_QNN_ONNX	float	Snapdragon® X Elite	2117.037 ms	951 - 951 MB	NPU
vision_backbone	PRECOMPILED_QNN_ONNX	float	Snapdragon® 8 Gen 3 Mobile	1670.675 ms	20 - 30 MB	NPU
vision_backbone	PRECOMPILED_QNN_ONNX	float	Qualcomm® Dragonwing™ IQ-8275	2459.713 ms	101 - 117 MB	NPU
vision_backbone	PRECOMPILED_QNN_ONNX	float	Qualcomm® Dragonwing™ IQ-9075	2415.958 ms	12 - 27 MB	NPU
vision_backbone	PRECOMPILED_QNN_ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	2117.037 ms	951 - 951 MB	NPU
vision_backbone	PRECOMPILED_QNN_ONNX	float	Qualcomm® Dragonwing™ Q-8750	1295.951 ms	8 - 19 MB	NPU
vision_backbone	QNN_CONTEXT_BINARY	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	1099.147 ms	5 - 13 MB	NPU
vision_backbone	QNN_CONTEXT_BINARY	float	Snapdragon® 8 Elite For Galaxy Mobile	1294.463 ms	0 - 14 MB	NPU
vision_backbone	QNN_CONTEXT_BINARY	float	Snapdragon® X2 Elite	1011.608 ms	12 - 12 MB	NPU
vision_backbone	QNN_CONTEXT_BINARY	float	Snapdragon® X Elite	2142.063 ms	12 - 12 MB	NPU
vision_backbone	QNN_CONTEXT_BINARY	float	Snapdragon® 8 Gen 3 Mobile	1638.502 ms	0 - 8 MB	NPU
vision_backbone	QNN_CONTEXT_BINARY	float	Qualcomm® SA8650P	2520.912 ms	1 - 11 MB	NPU
vision_backbone	QNN_CONTEXT_BINARY	float	Qualcomm® SA8255P	2520.912 ms	1 - 11 MB	NPU
vision_backbone	QNN_CONTEXT_BINARY	float	Qualcomm® Dragonwing™ IQ-9075	2413.638 ms	12 - 132 MB	NPU
vision_backbone	QNN_CONTEXT_BINARY	float	Qualcomm® Dragonwing™ IQ-X7181	2142.063 ms	12 - 12 MB	NPU
vision_backbone	QNN_CONTEXT_BINARY	float	Qualcomm® Dragonwing™ Q-8750	1294.463 ms	0 - 14 MB	NPU
vision_backbone	QNN_CONTEXT_BINARY	float	Qualcomm® SA8295P	2799.918 ms	11 - 21 MB	NPU

License

• The license for the original implementation of Segment-Anything-Model-3 can be found here.

References

• SAM 3: Segment Anything with Concepts
• 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.