EdgeTAM: Optimized for Qualcomm Devices

EdgeTAM is a lightweight and efficient version of SAM 2, optimized for on-device tracking and segmenting any object in videos and low memory usage by using a novel 2D Spatial Perceiver and RepViT backbone.

This is based on the implementation of EdgeTAM 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
ONNX w8a8 Universal QAIRT 2.42, ONNX Runtime 1.24.3 Download
QNN_DLC float Universal QAIRT 2.45 Download
QNN_DLC w8a8 Universal QAIRT 2.45 Download
TFLITE float Universal QAIRT 2.45 Download
TFLITE w8a8 Universal QAIRT 2.45 Download

For more device-specific assets and performance metrics, visit EdgeTAM 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 EdgeTAM on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

  • Model checkpoint: edgetam.pt
  • Input resolution (Encoder): 1024x1024
  • Number of parameters (encoder): 8.3M
  • Model size (encoder) (float): 33 MB
  • Number of parameters (decoder): 6.22M
  • Model size (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 3.154 ms 17 - 244 MB NPU
decoder ONNX float Snapdragon® 8 Elite Mobile 3.501 ms 2 - 241 MB NPU
decoder ONNX float Snapdragon® X2 Elite 3.277 ms 15 - 15 MB NPU
decoder ONNX float Snapdragon® X Elite 6.885 ms 16 - 16 MB NPU
decoder ONNX float Snapdragon® X Elite 6.885 ms 16 - 16 MB NPU
decoder ONNX float Snapdragon® 8 Gen 3 Mobile 4.729 ms 16 - 341 MB NPU
decoder ONNX float Qualcomm® QCS8550 (Proxy) 6.748 ms 16 - 24 MB NPU
decoder ONNX float Qualcomm® QCS9075 8.43 ms 16 - 19 MB NPU
decoder ONNX float Snapdragon® 8 Elite For Galaxy Mobile 3.501 ms 2 - 241 MB NPU
decoder ONNX w8a8 Snapdragon® 8 Elite Gen 5 Mobile 5.264 ms 4 - 268 MB NPU
decoder ONNX w8a8 Snapdragon® 8 Elite Mobile 6.731 ms 2 - 244 MB NPU
decoder ONNX w8a8 Snapdragon® X2 Elite 5.414 ms 4 - 4 MB NPU
decoder ONNX w8a8 Snapdragon® X Elite 12.489 ms 4 - 4 MB NPU
decoder ONNX w8a8 Snapdragon® X Elite 12.489 ms 4 - 4 MB NPU
decoder ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 9.194 ms 4 - 289 MB NPU
decoder ONNX w8a8 Qualcomm® QCS6490 126.268 ms 45 - 57 MB CPU
decoder ONNX w8a8 Qualcomm® QCS8550 (Proxy) 12.003 ms 2 - 13 MB NPU
decoder ONNX w8a8 Qualcomm® QCS9075 13.419 ms 4 - 7 MB NPU
decoder ONNX w8a8 Qualcomm® QCM6690 58.199 ms 44 - 67 MB CPU
decoder ONNX w8a8 Snapdragon® 8 Elite For Galaxy Mobile 6.731 ms 2 - 244 MB NPU
decoder ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 51.436 ms 41 - 60 MB CPU
decoder ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 51.436 ms 41 - 60 MB CPU
decoder QNN_DLC float Snapdragon® 8 Elite Gen 5 Mobile 2.607 ms 16 - 202 MB NPU
decoder QNN_DLC float Snapdragon® 8 Elite Mobile 2.878 ms 3 - 188 MB NPU
decoder QNN_DLC float Snapdragon® X2 Elite 3.086 ms 16 - 16 MB NPU
decoder QNN_DLC float Snapdragon® X Elite 5.589 ms 16 - 16 MB NPU
decoder QNN_DLC float Snapdragon® X Elite 5.589 ms 16 - 16 MB NPU
decoder QNN_DLC float Snapdragon® 8 Gen 3 Mobile 3.716 ms 4 - 215 MB NPU
decoder QNN_DLC float Qualcomm® QCS8275 (Proxy) 12.65 ms 16 - 201 MB NPU
decoder QNN_DLC float Qualcomm® QCS8550 (Proxy) 5.395 ms 16 - 18 MB NPU
decoder QNN_DLC float Qualcomm® SA8775P 6.479 ms 16 - 223 MB NPU
decoder QNN_DLC float Qualcomm® SA8775P 6.479 ms 16 - 223 MB NPU
decoder QNN_DLC float Qualcomm® SA8775P 6.479 ms 16 - 223 MB NPU
decoder QNN_DLC float Qualcomm® QCS9075 6.724 ms 16 - 34 MB NPU
decoder QNN_DLC float Qualcomm® QCS8450 (Proxy) 10.032 ms 4 - 220 MB NPU
decoder QNN_DLC float Qualcomm® SA7255P 12.65 ms 16 - 201 MB NPU
decoder QNN_DLC float Qualcomm® SA8295P 8.141 ms 0 - 193 MB NPU
decoder QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 2.878 ms 3 - 188 MB NPU
decoder QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 Mobile 0.894 ms 1 - 128 MB NPU
decoder QNN_DLC w8a8 Snapdragon® 8 Elite Mobile 1.02 ms 1 - 89 MB NPU
decoder QNN_DLC w8a8 Snapdragon® X2 Elite 1.372 ms 4 - 4 MB NPU
decoder QNN_DLC w8a8 Snapdragon® X Elite 2.279 ms 4 - 4 MB NPU
decoder QNN_DLC w8a8 Snapdragon® X Elite 2.279 ms 4 - 4 MB NPU
decoder QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 1.289 ms 0 - 109 MB NPU
decoder QNN_DLC w8a8 Qualcomm® QCS6490 8.481 ms 1 - 7 MB NPU
decoder QNN_DLC w8a8 Qualcomm® QCS8275 (Proxy) 3.887 ms 4 - 89 MB NPU
decoder QNN_DLC w8a8 Qualcomm® QCS8550 (Proxy) 1.976 ms 4 - 6 MB NPU
decoder QNN_DLC w8a8 Qualcomm® SA8775P 2.603 ms 4 - 90 MB NPU
decoder QNN_DLC w8a8 Qualcomm® SA8775P 2.603 ms 4 - 90 MB NPU
decoder QNN_DLC w8a8 Qualcomm® SA8775P 2.603 ms 4 - 90 MB NPU
decoder QNN_DLC w8a8 Qualcomm® QCS9075 2.405 ms 1 - 7 MB NPU
decoder QNN_DLC w8a8 Qualcomm® QCM6690 9.994 ms 4 - 272 MB NPU
decoder QNN_DLC w8a8 Qualcomm® QCS8450 (Proxy) 2.542 ms 0 - 109 MB NPU
decoder QNN_DLC w8a8 Qualcomm® SA7255P 3.887 ms 4 - 89 MB NPU
decoder QNN_DLC w8a8 Qualcomm® SA8295P 2.939 ms 4 - 88 MB NPU
decoder QNN_DLC w8a8 Snapdragon® 8 Elite For Galaxy Mobile 1.02 ms 1 - 89 MB NPU
decoder QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 2.703 ms 4 - 210 MB NPU
decoder QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 2.703 ms 4 - 210 MB NPU
decoder TFLITE float Snapdragon® 8 Elite Gen 5 Mobile 2.61 ms 0 - 188 MB NPU
decoder TFLITE float Snapdragon® 8 Elite Mobile 2.862 ms 0 - 182 MB NPU
decoder TFLITE float Snapdragon® 8 Gen 3 Mobile 3.733 ms 0 - 213 MB NPU
decoder TFLITE float Qualcomm® QCS8275 (Proxy) 12.623 ms 0 - 186 MB NPU
decoder TFLITE float Qualcomm® QCS8550 (Proxy) 5.336 ms 0 - 4 MB NPU
decoder TFLITE float Qualcomm® SA8775P 6.6 ms 0 - 188 MB NPU
decoder TFLITE float Qualcomm® SA8775P 6.6 ms 0 - 188 MB NPU
decoder TFLITE float Qualcomm® SA8775P 6.6 ms 0 - 188 MB NPU
decoder TFLITE float Qualcomm® QCS9075 6.793 ms 0 - 30 MB NPU
decoder TFLITE float Qualcomm® QCS8450 (Proxy) 10.071 ms 0 - 217 MB NPU
decoder TFLITE float Qualcomm® SA7255P 12.623 ms 0 - 186 MB NPU
decoder TFLITE float Qualcomm® SA8295P 8.261 ms 0 - 195 MB NPU
decoder TFLITE float Snapdragon® 8 Elite For Galaxy Mobile 2.862 ms 0 - 182 MB NPU
decoder TFLITE w8a8 Snapdragon® 8 Elite Gen 5 Mobile 3.092 ms 0 - 254 MB NPU
decoder TFLITE w8a8 Snapdragon® 8 Elite Mobile 3.514 ms 0 - 239 MB NPU
decoder TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 5.789 ms 0 - 264 MB NPU
decoder TFLITE w8a8 Qualcomm® QCS6490 31.722 ms 12 - 58 MB NPU
decoder TFLITE w8a8 Qualcomm® QCS8275 (Proxy) 13.634 ms 0 - 237 MB NPU
decoder TFLITE w8a8 Qualcomm® QCS8550 (Proxy) 7.72 ms 0 - 22 MB NPU
decoder TFLITE w8a8 Qualcomm® SA8775P 8.551 ms 0 - 240 MB NPU
decoder TFLITE w8a8 Qualcomm® SA8775P 8.551 ms 0 - 240 MB NPU
decoder TFLITE w8a8 Qualcomm® SA8775P 8.551 ms 0 - 240 MB NPU
decoder TFLITE w8a8 Qualcomm® QCS9075 8.799 ms 0 - 14 MB NPU
decoder TFLITE w8a8 Qualcomm® QCM6690 32.805 ms 12 - 210 MB NPU
decoder TFLITE w8a8 Qualcomm® QCS8450 (Proxy) 10.182 ms 0 - 246 MB NPU
decoder TFLITE w8a8 Qualcomm® SA7255P 13.634 ms 0 - 237 MB NPU
decoder TFLITE w8a8 Qualcomm® SA8295P 9.486 ms 0 - 240 MB NPU
decoder TFLITE w8a8 Snapdragon® 8 Elite For Galaxy Mobile 3.514 ms 0 - 239 MB NPU
decoder TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 14.221 ms 12 - 203 MB NPU
decoder TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 14.221 ms 12 - 203 MB NPU
encoder ONNX float Snapdragon® 8 Elite Gen 5 Mobile 7.813 ms 45 - 329 MB NPU
encoder ONNX float Snapdragon® 8 Elite Mobile 9.928 ms 33 - 308 MB NPU
encoder ONNX float Snapdragon® X2 Elite 8.901 ms 32 - 32 MB NPU
encoder ONNX float Snapdragon® X Elite 20.208 ms 34 - 34 MB NPU
encoder ONNX float Snapdragon® X Elite 20.208 ms 34 - 34 MB NPU
encoder ONNX float Snapdragon® 8 Gen 3 Mobile 15.205 ms 0 - 350 MB NPU
encoder ONNX float Qualcomm® QCS8550 (Proxy) 20.383 ms 28 - 44 MB NPU
encoder ONNX float Qualcomm® QCS9075 29.555 ms 12 - 26 MB NPU
encoder ONNX float Snapdragon® 8 Elite For Galaxy Mobile 9.928 ms 33 - 308 MB NPU
encoder ONNX w8a8 Snapdragon® 8 Elite Gen 5 Mobile 3.278 ms 5 - 279 MB NPU
encoder ONNX w8a8 Snapdragon® 8 Elite Mobile 4.425 ms 8 - 272 MB NPU
encoder ONNX w8a8 Snapdragon® X2 Elite 3.509 ms 18 - 18 MB NPU
encoder ONNX w8a8 Snapdragon® X Elite 8.889 ms 16 - 16 MB NPU
encoder ONNX w8a8 Snapdragon® X Elite 8.889 ms 16 - 16 MB NPU
encoder ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 5.465 ms 12 - 321 MB NPU
encoder ONNX w8a8 Qualcomm® QCS6490 487.693 ms 203 - 213 MB CPU
encoder ONNX w8a8 Qualcomm® QCS8550 (Proxy) 8.207 ms 11 - 18 MB NPU
encoder ONNX w8a8 Qualcomm® QCS9075 9.734 ms 12 - 14 MB NPU
encoder ONNX w8a8 Qualcomm® QCM6690 323.469 ms 208 - 219 MB CPU
encoder ONNX w8a8 Snapdragon® 8 Elite For Galaxy Mobile 4.425 ms 8 - 272 MB NPU
encoder ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 303.762 ms 139 - 151 MB CPU
encoder ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 303.762 ms 139 - 151 MB CPU
encoder QNN_DLC float Snapdragon® 8 Elite Gen 5 Mobile 8.85 ms 12 - 309 MB NPU
encoder QNN_DLC float Snapdragon® 8 Elite Mobile 11.847 ms 12 - 290 MB NPU
encoder QNN_DLC float Snapdragon® X2 Elite 10.503 ms 12 - 12 MB NPU
encoder QNN_DLC float Snapdragon® X Elite 24.284 ms 12 - 12 MB NPU
encoder QNN_DLC float Snapdragon® X Elite 24.284 ms 12 - 12 MB NPU
encoder QNN_DLC float Snapdragon® 8 Gen 3 Mobile 17.128 ms 0 - 363 MB NPU
encoder QNN_DLC float Qualcomm® QCS8275 (Proxy) 69.854 ms 0 - 280 MB NPU
encoder QNN_DLC float Qualcomm® QCS8550 (Proxy) 23.812 ms 12 - 26 MB NPU
encoder QNN_DLC float Qualcomm® SA8775P 28.067 ms 0 - 280 MB NPU
encoder QNN_DLC float Qualcomm® SA8775P 28.067 ms 0 - 280 MB NPU
encoder QNN_DLC float Qualcomm® SA8775P 28.067 ms 0 - 280 MB NPU
encoder QNN_DLC float Qualcomm® QCS9075 34.849 ms 12 - 42 MB NPU
encoder QNN_DLC float Qualcomm® QCS8450 (Proxy) 93.702 ms 0 - 434 MB NPU
encoder QNN_DLC float Qualcomm® SA7255P 69.854 ms 0 - 280 MB NPU
encoder QNN_DLC float Qualcomm® SA8295P 56.687 ms 0 - 358 MB NPU
encoder QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 11.847 ms 12 - 290 MB NPU
encoder QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 Mobile 4.207 ms 3 - 273 MB NPU
encoder QNN_DLC w8a8 Snapdragon® 8 Elite Mobile 5.36 ms 3 - 261 MB NPU
encoder QNN_DLC w8a8 Snapdragon® X2 Elite 4.845 ms 3 - 3 MB NPU
encoder QNN_DLC w8a8 Snapdragon® X Elite 12.114 ms 3 - 3 MB NPU
encoder QNN_DLC w8a8 Snapdragon® X Elite 12.114 ms 3 - 3 MB NPU
encoder QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 7.515 ms 3 - 313 MB NPU
encoder QNN_DLC w8a8 Qualcomm® QCS6490 3462.879 ms 3 - 12 MB NPU
encoder QNN_DLC w8a8 Qualcomm® QCS8275 (Proxy) 20.358 ms 3 - 267 MB NPU
encoder QNN_DLC w8a8 Qualcomm® QCS8550 (Proxy) 11.346 ms 3 - 5 MB NPU
encoder QNN_DLC w8a8 Qualcomm® SA8775P 12.039 ms 3 - 268 MB NPU
encoder QNN_DLC w8a8 Qualcomm® SA8775P 12.039 ms 3 - 268 MB NPU
encoder QNN_DLC w8a8 Qualcomm® SA8775P 12.039 ms 3 - 268 MB NPU
encoder QNN_DLC w8a8 Qualcomm® QCS9075 14.73 ms 1 - 10 MB NPU
encoder QNN_DLC w8a8 Qualcomm® QCM6690 58.734 ms 3 - 270 MB NPU
encoder QNN_DLC w8a8 Qualcomm® QCS8450 (Proxy) 1940.561 ms 3 - 209 MB NPU
encoder QNN_DLC w8a8 Qualcomm® SA7255P 20.358 ms 3 - 267 MB NPU
encoder QNN_DLC w8a8 Qualcomm® SA8295P 1982.304 ms 3 - 165 MB NPU
encoder QNN_DLC w8a8 Snapdragon® 8 Elite For Galaxy Mobile 5.36 ms 3 - 261 MB NPU
encoder QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 12.273 ms 3 - 300 MB NPU
encoder QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 12.273 ms 3 - 300 MB NPU
encoder TFLITE float Snapdragon® 8 Elite Gen 5 Mobile 8.655 ms 16 - 321 MB NPU
encoder TFLITE float Snapdragon® 8 Elite Mobile 11.437 ms 14 - 301 MB NPU
encoder TFLITE float Snapdragon® 8 Gen 3 Mobile 16.567 ms 15 - 366 MB NPU
encoder TFLITE float Qualcomm® QCS8275 (Proxy) 68.116 ms 16 - 289 MB NPU
encoder TFLITE float Qualcomm® QCS8550 (Proxy) 23.013 ms 16 - 18 MB NPU
encoder TFLITE float Qualcomm® SA8775P 27.372 ms 16 - 287 MB NPU
encoder TFLITE float Qualcomm® SA8775P 27.372 ms 16 - 287 MB NPU
encoder TFLITE float Qualcomm® SA8775P 27.372 ms 16 - 287 MB NPU
encoder TFLITE float Qualcomm® QCS9075 33.739 ms 0 - 42 MB NPU
encoder TFLITE float Qualcomm® QCS8450 (Proxy) 97.078 ms 16 - 438 MB NPU
encoder TFLITE float Qualcomm® SA7255P 68.116 ms 16 - 289 MB NPU
encoder TFLITE float Qualcomm® SA8295P 53.478 ms 16 - 366 MB NPU
encoder TFLITE float Snapdragon® 8 Elite For Galaxy Mobile 11.437 ms 14 - 301 MB NPU
encoder TFLITE w8a8 Snapdragon® 8 Elite Gen 5 Mobile 4.137 ms 4 - 263 MB NPU
encoder TFLITE w8a8 Snapdragon® 8 Elite Mobile 3.022 ms 4 - 255 MB NPU
encoder TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 4.772 ms 2 - 304 MB NPU
encoder TFLITE w8a8 Qualcomm® QCS6490 3194.369 ms 0 - 22 MB NPU
encoder TFLITE w8a8 Qualcomm® QCS8275 (Proxy) 13.741 ms 4 - 261 MB NPU
encoder TFLITE w8a8 Qualcomm® QCS8550 (Proxy) 7.025 ms 4 - 6 MB NPU
encoder TFLITE w8a8 Qualcomm® SA8775P 7.821 ms 4 - 261 MB NPU
encoder TFLITE w8a8 Qualcomm® SA8775P 7.821 ms 4 - 261 MB NPU
encoder TFLITE w8a8 Qualcomm® SA8775P 7.821 ms 4 - 261 MB NPU
encoder TFLITE w8a8 Qualcomm® QCS9075 7.365 ms 2 - 17 MB NPU
encoder TFLITE w8a8 Qualcomm® QCM6690 56.598 ms 2 - 234 MB NPU
encoder TFLITE w8a8 Qualcomm® QCS8450 (Proxy) 2262.099 ms 1 - 329 MB NPU
encoder TFLITE w8a8 Qualcomm® SA7255P 13.741 ms 4 - 261 MB NPU
encoder TFLITE w8a8 Qualcomm® SA8295P 2137.627 ms 0 - 281 MB NPU
encoder TFLITE w8a8 Snapdragon® 8 Elite For Galaxy Mobile 3.022 ms 4 - 255 MB NPU
encoder TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 8.163 ms 4 - 261 MB NPU
encoder TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 8.163 ms 4 - 261 MB NPU

License

  • The license for the original implementation of EdgeTAM 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/EdgeTAM