Segment-Anything-Model-2: Optimized for Qualcomm Devices

SAM 2, the successor to Meta's Segment Anything Model (SAM), is a cutting-edge tool designed for comprehensive object segmentation in both images and videos. It excels in handling complex visual data through a unified, promptable model architecture that supports real-time processing and zero-shot generalization.

This is based on the implementation of Segment-Anything-Model-2 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 w8a8 Universal QAIRT 2.50, ONNX Runtime 1.27.1 Download
QNN_DLC float Universal QAIRT 2.50 Download
QNN_DLC w8a8 Universal QAIRT 2.50 Download
TFLITE w8a8 Universal QAIRT 2.50 Download

For more device-specific assets and performance metrics, visit Segment-Anything-Model-2 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 Segment-Anything-Model-2 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

  • Input resolution: 720p (720x1280)
  • Model checkpoint: sam2.1_hiera_t
  • Model size (decoder) (float): 23.7 MB
  • Model size (encoder) (float): 128 MB
  • Number of parameters (decoder): 6.22M
  • Number of parameters (encoder): 33.5M

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
decoder ONNX float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 3.275 ms 2 - 221 MB NPU
decoder ONNX float Snapdragon® X2 Elite 3.226 ms 16 - 16 MB NPU
decoder ONNX float Snapdragon® X Elite 6.798 ms 16 - 16 MB NPU
decoder ONNX float Snapdragon® 8 Gen 3 Mobile 4.614 ms 17 - 262 MB NPU
decoder ONNX float Snapdragon® 8 Gen 1 Mobile 10.943 ms 17 - 255 MB NPU
decoder ONNX float Qualcomm® Dragonwing™ IQ-8275 7.402 ms 16 - 20 MB NPU
decoder ONNX float Qualcomm® Dragonwing™ QCS8550 (Proxy) 6.674 ms 16 - 23 MB NPU
decoder ONNX float Qualcomm® QCS8450 10.943 ms 17 - 255 MB NPU
decoder ONNX float Qualcomm® Dragonwing™ IQ-9075 7.718 ms 16 - 19 MB NPU
decoder ONNX float Qualcomm® Dragonwing™ IQ-X7181 6.798 ms 16 - 16 MB NPU
decoder ONNX w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 1.418 ms 0 - 141 MB NPU
decoder ONNX w8a8 Snapdragon® 8 Elite For Galaxy Mobile 1.582 ms 0 - 121 MB NPU
decoder ONNX w8a8 Snapdragon® X2 Elite 1.485 ms 4 - 4 MB NPU
decoder ONNX w8a8 Snapdragon® X Elite 2.895 ms 4 - 4 MB NPU
decoder ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 1.904 ms 4 - 145 MB NPU
decoder ONNX w8a8 Snapdragon® 8 Gen 1 Mobile 3.525 ms 4 - 146 MB NPU
decoder ONNX w8a8 Qualcomm® Dragonwing™ QCS6490 9.654 ms 4 - 7 MB NPU
decoder ONNX w8a8 Qualcomm® Dragonwing™ IQ-8275 2.991 ms 4 - 8 MB NPU
decoder ONNX w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 2.798 ms 2 - 8 MB NPU
decoder ONNX w8a8 Qualcomm® QCS8450 3.525 ms 4 - 146 MB NPU
decoder ONNX w8a8 Qualcomm® Dragonwing™ IQ-9075 3.359 ms 4 - 7 MB NPU
decoder ONNX w8a8 Qualcomm® Dragonwing™ IQ-X7181 2.895 ms 4 - 4 MB NPU
decoder ONNX w8a8 Qualcomm® Dragonwing™ Q-6690 14.83 ms 3 - 233 MB NPU
decoder ONNX w8a8 Qualcomm® Dragonwing™ Q-7790 3.585 ms 4 - 238 MB NPU
decoder ONNX w8a8 Qualcomm® Dragonwing™ Q-8750 1.582 ms 0 - 121 MB NPU
decoder ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 3.585 ms 4 - 238 MB NPU
decoder QNN_DLC float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 2.43 ms 4 - 192 MB NPU
decoder QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 2.811 ms 4 - 184 MB NPU
decoder QNN_DLC float Snapdragon® X2 Elite 2.954 ms 16 - 16 MB NPU
decoder QNN_DLC float Snapdragon® X Elite 5.641 ms 16 - 16 MB NPU
decoder QNN_DLC float Snapdragon® 8 Gen 3 Mobile 3.653 ms 4 - 211 MB NPU
decoder QNN_DLC float Snapdragon® 8 Gen 1 Mobile 7.587 ms 4 - 209 MB NPU
decoder QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 6.09 ms 16 - 35 MB NPU
decoder QNN_DLC float Qualcomm® QCS8450 7.587 ms 4 - 209 MB NPU
decoder QNN_DLC float Qualcomm® Dragonwing™ IQ-9075 6.424 ms 16 - 34 MB NPU
decoder QNN_DLC float Qualcomm® Dragonwing™ IQ-X7181 5.641 ms 16 - 16 MB NPU
decoder QNN_DLC float Qualcomm® Dragonwing™ Q-8750 2.811 ms 4 - 184 MB NPU
decoder QNN_DLC float Qualcomm® SA8295P 8.11 ms 0 - 186 MB NPU
decoder QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 0.86 ms 1 - 111 MB NPU
decoder QNN_DLC w8a8 Snapdragon® 8 Elite For Galaxy Mobile 0.962 ms 1 - 94 MB NPU
decoder QNN_DLC w8a8 Snapdragon® X2 Elite 1.283 ms 4 - 4 MB NPU
decoder QNN_DLC w8a8 Snapdragon® X Elite 2.294 ms 4 - 4 MB NPU
decoder QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 1.288 ms 0 - 107 MB NPU
decoder QNN_DLC w8a8 Snapdragon® 8 Gen 1 Mobile 2.396 ms 0 - 111 MB NPU
decoder QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS6490 6.513 ms 1 - 7 MB NPU
decoder QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-8275 2.145 ms 1 - 8 MB NPU
decoder QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 1.93 ms 4 - 6 MB NPU
decoder QNN_DLC w8a8 Qualcomm® SA8650P 2.559 ms 4 - 89 MB NPU
decoder QNN_DLC w8a8 Qualcomm® SA8255P 2.559 ms 4 - 89 MB NPU
decoder QNN_DLC w8a8 Qualcomm® QCS8450 2.396 ms 0 - 111 MB NPU
decoder QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-9075 2.392 ms 1 - 7 MB NPU
decoder QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-X7181 2.294 ms 4 - 4 MB NPU
decoder QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-6690 9.778 ms 4 - 205 MB NPU
decoder QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-7790 2.496 ms 4 - 207 MB NPU
decoder QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-8750 0.962 ms 1 - 94 MB NPU
decoder QNN_DLC w8a8 Qualcomm® SA8295P 2.821 ms 4 - 90 MB NPU
decoder QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 2.496 ms 4 - 207 MB NPU
decoder TFLITE float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 2.468 ms 0 - 181 MB NPU
decoder TFLITE float Snapdragon® 8 Gen 3 Mobile 3.673 ms 0 - 215 MB NPU
decoder TFLITE float Snapdragon® 8 Gen 1 Mobile 7.645 ms 0 - 208 MB NPU
decoder TFLITE float Qualcomm® Dragonwing™ IQ-8275 6.124 ms 0 - 31 MB NPU
decoder TFLITE float Qualcomm® Dragonwing™ QCS8550 (Proxy) 5.387 ms 0 - 8 MB NPU
decoder TFLITE float Qualcomm® QCS8450 7.645 ms 0 - 208 MB NPU
decoder TFLITE float Qualcomm® Dragonwing™ IQ-9075 6.373 ms 0 - 31 MB NPU
decoder TFLITE float Qualcomm® SA8295P 8.208 ms 0 - 186 MB NPU
decoder TFLITE w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 2.048 ms 0 - 231 MB NPU
decoder TFLITE w8a8 Snapdragon® 8 Elite For Galaxy Mobile 3.553 ms 0 - 205 MB NPU
decoder TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 5.979 ms 0 - 250 MB NPU
decoder TFLITE w8a8 Snapdragon® 8 Gen 1 Mobile 9.012 ms 0 - 233 MB NPU
decoder TFLITE w8a8 Qualcomm® Dragonwing™ QCS6490 31.528 ms 12 - 59 MB NPU
decoder TFLITE w8a8 Qualcomm® Dragonwing™ IQ-8275 7.966 ms 0 - 15 MB NPU
decoder TFLITE w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 7.908 ms 0 - 23 MB NPU
decoder TFLITE w8a8 Qualcomm® SA8650P 8.613 ms 0 - 240 MB NPU
decoder TFLITE w8a8 Qualcomm® SA8255P 8.613 ms 0 - 240 MB NPU
decoder TFLITE w8a8 Qualcomm® QCS8450 9.012 ms 0 - 233 MB NPU
decoder TFLITE w8a8 Qualcomm® Dragonwing™ IQ-9075 9.378 ms 0 - 14 MB NPU
decoder TFLITE w8a8 Qualcomm® Dragonwing™ Q-6690 31.958 ms 12 - 191 MB NPU
decoder TFLITE w8a8 Qualcomm® Dragonwing™ Q-7790 13.8 ms 12 - 209 MB NPU
decoder TFLITE w8a8 Qualcomm® Dragonwing™ Q-8750 3.553 ms 0 - 205 MB NPU
decoder TFLITE w8a8 Qualcomm® SA8295P 10.415 ms 0 - 210 MB NPU
decoder TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 13.8 ms 12 - 209 MB NPU
encoder ONNX float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 66.206 ms 44 - 1563 MB NPU
encoder ONNX float Snapdragon® X2 Elite 74.683 ms 70 - 70 MB NPU
encoder ONNX float Snapdragon® X Elite 158.565 ms 72 - 72 MB NPU
encoder ONNX float Snapdragon® 8 Gen 3 Mobile 119.051 ms 43 - 1969 MB NPU
encoder ONNX float Snapdragon® 8 Gen 1 Mobile 249.665 ms 55 - 1909 MB NPU
encoder ONNX float Qualcomm® Dragonwing™ IQ-8275 159.909 ms 12 - 28 MB NPU
encoder ONNX float Qualcomm® Dragonwing™ QCS8550 (Proxy) 160.289 ms 8 - 94 MB NPU
encoder ONNX float Qualcomm® QCS8450 249.665 ms 55 - 1909 MB NPU
encoder ONNX float Qualcomm® Dragonwing™ IQ-9075 174.036 ms 12 - 27 MB NPU
encoder ONNX float Qualcomm® Dragonwing™ IQ-X7181 158.565 ms 72 - 72 MB NPU
encoder ONNX w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 38.634 ms 11 - 1835 MB NPU
encoder ONNX w8a8 Snapdragon® 8 Elite For Galaxy Mobile 38.071 ms 12 - 1642 MB NPU
encoder ONNX w8a8 Snapdragon® X2 Elite 38.409 ms 18 - 18 MB NPU
encoder ONNX w8a8 Snapdragon® X Elite 73.255 ms 46 - 46 MB NPU
encoder ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 51.03 ms 15 - 1924 MB NPU
encoder ONNX w8a8 Snapdragon® 8 Gen 1 Mobile 96.611 ms 11 - 2357 MB NPU
encoder ONNX w8a8 Qualcomm® Dragonwing™ QCS6490 600.746 ms 7 - 13 MB NPU
encoder ONNX w8a8 Qualcomm® Dragonwing™ IQ-8275 62.93 ms 10 - 17 MB NPU
encoder ONNX w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 71.19 ms 12 - 16 MB NPU
encoder ONNX w8a8 Qualcomm® QCS8450 96.611 ms 11 - 2357 MB NPU
encoder ONNX w8a8 Qualcomm® Dragonwing™ IQ-9075 72.652 ms 10 - 17 MB NPU
encoder ONNX w8a8 Qualcomm® Dragonwing™ IQ-X7181 73.255 ms 46 - 46 MB NPU
encoder ONNX w8a8 Qualcomm® Dragonwing™ Q-6690 1442.123 ms 3 - 1969 MB NPU
encoder ONNX w8a8 Qualcomm® Dragonwing™ Q-7790 93.904 ms 4 - 1688 MB NPU
encoder ONNX w8a8 Qualcomm® Dragonwing™ Q-8750 38.071 ms 12 - 1642 MB NPU
encoder ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 93.904 ms 4 - 1688 MB NPU
encoder QNN_DLC float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 67.504 ms 0 - 1541 MB NPU
encoder QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 103.744 ms 12 - 1564 MB NPU
encoder QNN_DLC float Snapdragon® X2 Elite 76.104 ms 12 - 12 MB NPU
encoder QNN_DLC float Snapdragon® X Elite 165.679 ms 12 - 12 MB NPU
encoder QNN_DLC float Snapdragon® 8 Gen 3 Mobile 125.003 ms 160 - 4535 MB NPU
encoder QNN_DLC float Snapdragon® 8 Gen 1 Mobile 260.542 ms 12 - 1898 MB NPU
encoder QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 171.281 ms 12 - 47 MB NPU
encoder QNN_DLC float Qualcomm® QCS8450 260.542 ms 12 - 1898 MB NPU
encoder QNN_DLC float Qualcomm® Dragonwing™ IQ-9075 179.794 ms 14 - 48 MB NPU
encoder QNN_DLC float Qualcomm® Dragonwing™ IQ-X7181 165.679 ms 12 - 12 MB NPU
encoder QNN_DLC float Qualcomm® Dragonwing™ Q-8750 103.744 ms 12 - 1564 MB NPU
encoder QNN_DLC float Qualcomm® SA8295P 249.436 ms 0 - 1422 MB NPU
encoder QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 39.985 ms 3 - 1818 MB NPU
encoder QNN_DLC w8a8 Snapdragon® 8 Elite For Galaxy Mobile 52.181 ms 3 - 1706 MB NPU
encoder QNN_DLC w8a8 Snapdragon® X2 Elite 39.448 ms 3 - 3 MB NPU
encoder QNN_DLC w8a8 Snapdragon® X Elite 79.605 ms 3 - 3 MB NPU
encoder QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 53.585 ms 3 - 1893 MB NPU
encoder QNN_DLC w8a8 Snapdragon® 8 Gen 1 Mobile 106.671 ms 3 - 2329 MB NPU
encoder QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS6490 603.538 ms 3 - 13 MB NPU
encoder QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-8275 67.544 ms 3 - 14 MB NPU
encoder QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 76.656 ms 3 - 949 MB NPU
encoder QNN_DLC w8a8 Qualcomm® SA8650P 81.826 ms 3 - 1581 MB NPU
encoder QNN_DLC w8a8 Qualcomm® SA8255P 81.826 ms 3 - 1581 MB NPU
encoder QNN_DLC w8a8 Qualcomm® QCS8450 106.671 ms 3 - 2329 MB NPU
encoder QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-9075 78.104 ms 2 - 12 MB NPU
encoder QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-X7181 79.605 ms 3 - 3 MB NPU
encoder QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-6690 1580.151 ms 3 - 1960 MB NPU
encoder QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-7790 104.206 ms 3 - 2221 MB NPU
encoder QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-8750 52.181 ms 3 - 1706 MB NPU
encoder QNN_DLC w8a8 Qualcomm® SA8295P 92.472 ms 0 - 1598 MB NPU
encoder QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 104.206 ms 3 - 2221 MB NPU
encoder TFLITE float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 67.529 ms 7 - 1596 MB NPU
encoder TFLITE float Snapdragon® 8 Gen 3 Mobile 125.102 ms 20 - 2023 MB NPU
encoder TFLITE float Snapdragon® 8 Gen 1 Mobile 265.058 ms 21 - 1942 MB NPU
encoder TFLITE float Qualcomm® Dragonwing™ IQ-8275 171.938 ms 0 - 119 MB NPU
encoder TFLITE float Qualcomm® Dragonwing™ QCS8550 (Proxy) 171.625 ms 20 - 24 MB NPU
encoder TFLITE float Qualcomm® QCS8450 265.058 ms 21 - 1942 MB NPU
encoder TFLITE float Qualcomm® Dragonwing™ IQ-9075 179.985 ms 0 - 118 MB NPU
encoder TFLITE float Qualcomm® SA8295P 248.78 ms 20 - 1494 MB NPU
encoder TFLITE w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 34.806 ms 4 - 1804 MB NPU
encoder TFLITE w8a8 Snapdragon® 8 Elite For Galaxy Mobile 42.189 ms 3 - 1694 MB NPU
encoder TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 56.179 ms 5 - 1969 MB NPU
encoder TFLITE w8a8 Snapdragon® 8 Gen 1 Mobile 109.849 ms 5 - 2220 MB NPU
encoder TFLITE w8a8 Qualcomm® Dragonwing™ QCS6490 1235.538 ms 60 - 283 MB NPU
encoder TFLITE w8a8 Qualcomm® Dragonwing™ IQ-8275 71.251 ms 5 - 65 MB NPU
encoder TFLITE w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 80.076 ms 5 - 9 MB NPU
encoder TFLITE w8a8 Qualcomm® SA8650P 98.068 ms 6 - 1877 MB NPU
encoder TFLITE w8a8 Qualcomm® SA8255P 98.068 ms 6 - 1877 MB NPU
encoder TFLITE w8a8 Qualcomm® QCS8450 109.849 ms 5 - 2220 MB NPU
encoder TFLITE w8a8 Qualcomm® Dragonwing™ IQ-9075 80.516 ms 2 - 61 MB NPU
encoder TFLITE w8a8 Qualcomm® Dragonwing™ Q-6690 2108.452 ms 60 - 1036 MB NPU
encoder TFLITE w8a8 Qualcomm® Dragonwing™ Q-7790 306.757 ms 60 - 1008 MB NPU
encoder TFLITE w8a8 Qualcomm® Dragonwing™ Q-8750 42.189 ms 3 - 1694 MB NPU
encoder TFLITE w8a8 Qualcomm® SA8295P 98.748 ms 5 - 1692 MB NPU
encoder TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 306.757 ms 60 - 1008 MB NPU

License

  • The license for the original implementation of Segment-Anything-Model-2 can be found here.

References

Community

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Paper for qualcomm/Segment-Anything-Model-2