FastSam-S: Optimized for Qualcomm Devices

The Fast Segment Anything Model (FastSAM) is a novel, real-time CNN-based solution for the Segment Anything task. This task is designed to segment any object within an image based on various possible user interaction prompts. The model performs competitively despite significantly reduced computation, making it a practical choice for a variety of vision tasks.

This is based on the implementation of FastSam-S 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 FastSam-S on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

• Inference latency: RealTime
• Input resolution: 640x640
• Model checkpoint: fastsam-s.pt
• Model size (float): 45.1 MB
• Number of parameters: 11.8M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
FastSam-S	ONNX	float	Snapdragon® X2 Elite	4.587 ms	17 - 17 MB	NPU
FastSam-S	ONNX	float	Snapdragon® X Elite	8.522 ms	21 - 21 MB	NPU
FastSam-S	ONNX	float	Snapdragon® 8 Gen 3 Mobile	6.004 ms	18 - 249 MB	NPU
FastSam-S	ONNX	float	Snapdragon® 8 Gen 1 Mobile	16.632 ms	17 - 238 MB	NPU
FastSam-S	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	13.173 ms	15 - 19 MB	NPU
FastSam-S	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	8.161 ms	0 - 25 MB	NPU
FastSam-S	ONNX	float	Qualcomm® QCS8450	16.632 ms	17 - 238 MB	NPU
FastSam-S	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	11.988 ms	15 - 18 MB	NPU
FastSam-S	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	8.522 ms	21 - 21 MB	NPU
FastSam-S	ONNX	float	Qualcomm® Dragonwing™ Q-8750	4.686 ms	12 - 213 MB	NPU
FastSam-S	ONNX	float	Snapdragon® 8 Elite Mobile	4.686 ms	12 - 213 MB	NPU
FastSam-S	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	3.641 ms	3 - 212 MB	NPU
FastSam-S	ONNX	w8a16	Snapdragon® X2 Elite	3.132 ms	9 - 9 MB	NPU
FastSam-S	ONNX	w8a16	Snapdragon® X Elite	6.996 ms	10 - 10 MB	NPU
FastSam-S	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	4.381 ms	8 - 264 MB	NPU
FastSam-S	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	11.327 ms	8 - 266 MB	NPU
FastSam-S	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	28.223 ms	8 - 10 MB	NPU
FastSam-S	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-8275	6.444 ms	8 - 11 MB	NPU
FastSam-S	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.657 ms	0 - 14 MB	NPU
FastSam-S	ONNX	w8a16	Qualcomm® QCS8450	11.327 ms	8 - 266 MB	NPU
FastSam-S	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	7.028 ms	8 - 10 MB	NPU
FastSam-S	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	6.996 ms	10 - 10 MB	NPU
FastSam-S	ONNX	w8a16	Qualcomm® Dragonwing™ Q-6690	71.034 ms	8 - 266 MB	NPU
FastSam-S	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	9.229 ms	8 - 228 MB	NPU
FastSam-S	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	3.285 ms	1 - 217 MB	NPU
FastSam-S	ONNX	w8a16	Snapdragon® 8 Elite Mobile	3.285 ms	1 - 217 MB	NPU
FastSam-S	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	2.697 ms	1 - 226 MB	NPU
FastSam-S	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	9.229 ms	8 - 228 MB	NPU
FastSam-S	QNN_DLC	float	Snapdragon® X2 Elite	4.514 ms	5 - 5 MB	NPU
FastSam-S	QNN_DLC	float	Snapdragon® X Elite	8.483 ms	5 - 5 MB	NPU
FastSam-S	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	5.871 ms	5 - 215 MB	NPU
FastSam-S	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	16.119 ms	5 - 212 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	12.806 ms	5 - 16 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	39.184 ms	1 - 187 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	7.726 ms	5 - 7 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® SA8775P	11.578 ms	0 - 180 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® SA8650P	11.578 ms	0 - 180 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® SA8255P	11.578 ms	0 - 180 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® QCS8450	16.119 ms	5 - 212 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	11.387 ms	5 - 15 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	8.483 ms	5 - 5 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	4.488 ms	5 - 195 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® SA7255P	39.184 ms	1 - 187 MB	NPU
FastSam-S	QNN_DLC	float	Qualcomm® SA8295P	13.787 ms	0 - 178 MB	NPU
FastSam-S	QNN_DLC	float	Snapdragon® 8 Elite Mobile	4.488 ms	5 - 195 MB	NPU
FastSam-S	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	3.259 ms	5 - 197 MB	NPU
FastSam-S	QNN_DLC	w8a16	Snapdragon® X2 Elite	2.837 ms	2 - 2 MB	NPU
FastSam-S	QNN_DLC	w8a16	Snapdragon® X Elite	6.087 ms	2 - 2 MB	NPU
FastSam-S	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	3.772 ms	2 - 234 MB	NPU
FastSam-S	QNN_DLC	w8a16	Snapdragon® 8 Gen 1 Mobile	8.208 ms	2 - 233 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS6490	23.998 ms	2 - 8 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	5.559 ms	1 - 8 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	13.388 ms	0 - 193 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	5.493 ms	2 - 177 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® SA8775P	5.995 ms	1 - 197 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® SA8650P	5.995 ms	1 - 197 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® SA8255P	5.995 ms	1 - 197 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® QCS8450	8.208 ms	2 - 233 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	5.756 ms	3 - 9 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	6.087 ms	2 - 2 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	65.977 ms	2 - 203 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	7.591 ms	0 - 196 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	2.659 ms	2 - 200 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® SA7255P	13.388 ms	0 - 193 MB	NPU
FastSam-S	QNN_DLC	w8a16	Qualcomm® SA8295P	7.954 ms	0 - 191 MB	NPU
FastSam-S	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	2.659 ms	2 - 200 MB	NPU
FastSam-S	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	2.108 ms	2 - 204 MB	NPU
FastSam-S	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	7.591 ms	0 - 196 MB	NPU
FastSam-S	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	5.17 ms	4 - 118 MB	NPU
FastSam-S	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	14.048 ms	5 - 234 MB	NPU
FastSam-S	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	11.96 ms	4 - 40 MB	NPU
FastSam-S	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	37.807 ms	4 - 86 MB	NPU
FastSam-S	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.927 ms	4 - 7 MB	NPU
FastSam-S	TFLITE	float	Qualcomm® SA8775P	10.729 ms	0 - 86 MB	NPU
FastSam-S	TFLITE	float	Qualcomm® SA8650P	10.729 ms	0 - 86 MB	NPU
FastSam-S	TFLITE	float	Qualcomm® SA8255P	10.729 ms	0 - 86 MB	NPU
FastSam-S	TFLITE	float	Qualcomm® QCS8450	14.048 ms	5 - 234 MB	NPU
FastSam-S	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	10.443 ms	4 - 39 MB	NPU
FastSam-S	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	3.945 ms	0 - 97 MB	NPU
FastSam-S	TFLITE	float	Qualcomm® SA7255P	37.807 ms	4 - 86 MB	NPU
FastSam-S	TFLITE	float	Qualcomm® SA8295P	13.075 ms	4 - 195 MB	NPU
FastSam-S	TFLITE	float	Snapdragon® 8 Elite Mobile	3.945 ms	0 - 97 MB	NPU
FastSam-S	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	3.122 ms	0 - 90 MB	NPU

License

• The license for the original implementation of FastSam-S can be found here.

References

• Fast 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.