ResNet-Mixed-Convolution: Optimized for Qualcomm Devices

ResNet Mixed Convolutions is a network with a mixture of 2D and 3D convolutions used for video understanding.

This is based on the implementation of ResNet-Mixed-Convolution 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 2.45, ONNX Runtime 1.27.1 Download
ONNX w8a16 Universal QAIRT 2.45, ONNX Runtime 1.27.1 Download
ONNX w8a8 Universal QAIRT 2.45, ONNX Runtime 1.27.1 Download
QNN_DLC float Universal QAIRT 2.45 Download
QNN_DLC w8a16 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 ResNet-Mixed-Convolution 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 ResNet-Mixed-Convolution on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.video_classification

Model Stats:

  • Model checkpoint: Kinetics-400
  • Input resolution: 112x112
  • Number of parameters: 11.7M
  • Model size (float): 44.6 MB
  • Model size (w8a16): 11.5 MB

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
ResNet-Mixed-Convolution ONNX float Snapdragon® X2 Elite 37.751 ms 12 - 12 MB NPU
ResNet-Mixed-Convolution ONNX float Snapdragon® X Elite 70.124 ms 18 - 18 MB NPU
ResNet-Mixed-Convolution ONNX float Snapdragon® 8 Gen 3 Mobile 50.297 ms 12 - 720 MB NPU
ResNet-Mixed-Convolution ONNX float Snapdragon® 8 Gen 1 Mobile 132.296 ms 13 - 523 MB NPU
ResNet-Mixed-Convolution ONNX float Qualcomm® Dragonwing™ QCS8550 (Proxy) 69.943 ms 12 - 15 MB NPU
ResNet-Mixed-Convolution ONNX float Qualcomm® QCS8450 132.296 ms 13 - 523 MB NPU
ResNet-Mixed-Convolution ONNX float Qualcomm® Dragonwing™ IQ-9075 126.957 ms 11 - 14 MB NPU
ResNet-Mixed-Convolution ONNX float Qualcomm® Dragonwing™ IQ-X7181 70.124 ms 18 - 18 MB NPU
ResNet-Mixed-Convolution ONNX float Qualcomm® Dragonwing™ Q-8750 42.456 ms 1 - 540 MB NPU
ResNet-Mixed-Convolution ONNX float Snapdragon® 8 Elite Mobile 42.456 ms 1 - 540 MB NPU
ResNet-Mixed-Convolution ONNX float Snapdragon® 8 Elite Gen 5 Mobile 30.714 ms 0 - 566 MB NPU
ResNet-Mixed-Convolution ONNX w8a16 Snapdragon® X2 Elite 36.69 ms 7 - 7 MB NPU
ResNet-Mixed-Convolution ONNX w8a16 Snapdragon® X Elite 45.197 ms 11 - 11 MB NPU
ResNet-Mixed-Convolution ONNX w8a16 Snapdragon® 8 Gen 3 Mobile 33.062 ms 6 - 583 MB NPU
ResNet-Mixed-Convolution ONNX w8a16 Snapdragon® 8 Gen 1 Mobile 78.934 ms 8 - 585 MB NPU
ResNet-Mixed-Convolution ONNX w8a16 Qualcomm® Dragonwing™ QCS8550 (Proxy) 44.04 ms 0 - 13 MB NPU
ResNet-Mixed-Convolution ONNX w8a16 Qualcomm® QCS8450 78.934 ms 8 - 585 MB NPU
ResNet-Mixed-Convolution ONNX w8a16 Qualcomm® Dragonwing™ IQ-9075 49.314 ms 6 - 9 MB NPU
ResNet-Mixed-Convolution ONNX w8a16 Qualcomm® Dragonwing™ IQ-X7181 45.197 ms 11 - 11 MB NPU
ResNet-Mixed-Convolution ONNX w8a16 Qualcomm® Dragonwing™ Q-8750 26.496 ms 2 - 383 MB NPU
ResNet-Mixed-Convolution ONNX w8a16 Snapdragon® 8 Elite Mobile 26.496 ms 2 - 383 MB NPU
ResNet-Mixed-Convolution ONNX w8a16 Snapdragon® 8 Elite Gen 5 Mobile 19.734 ms 1 - 411 MB NPU
ResNet-Mixed-Convolution ONNX w8a8 Snapdragon® X2 Elite 11.541 ms 3 - 3 MB NPU
ResNet-Mixed-Convolution ONNX w8a8 Snapdragon® X Elite 24.403 ms 12 - 12 MB NPU
ResNet-Mixed-Convolution ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 16.928 ms 3 - 444 MB NPU
ResNet-Mixed-Convolution ONNX w8a8 Snapdragon® 8 Gen 1 Mobile 34.095 ms 3 - 442 MB NPU
ResNet-Mixed-Convolution ONNX w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 23.705 ms 0 - 15 MB NPU
ResNet-Mixed-Convolution ONNX w8a8 Qualcomm® QCS8450 34.095 ms 3 - 442 MB NPU
ResNet-Mixed-Convolution ONNX w8a8 Qualcomm® Dragonwing™ IQ-9075 23.447 ms 3 - 6 MB NPU
ResNet-Mixed-Convolution ONNX w8a8 Qualcomm® Dragonwing™ IQ-X7181 24.403 ms 12 - 12 MB NPU
ResNet-Mixed-Convolution ONNX w8a8 Qualcomm® Dragonwing™ Q-8750 14.241 ms 0 - 337 MB NPU
ResNet-Mixed-Convolution ONNX w8a8 Snapdragon® 8 Elite Mobile 14.241 ms 0 - 337 MB NPU
ResNet-Mixed-Convolution ONNX w8a8 Snapdragon® 8 Elite Gen 5 Mobile 9.363 ms 0 - 345 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Snapdragon® X2 Elite 38.045 ms 12 - 12 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Snapdragon® X Elite 71.038 ms 12 - 12 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Snapdragon® 8 Gen 3 Mobile 50.664 ms 12 - 717 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Snapdragon® 8 Gen 1 Mobile 133.815 ms 3 - 514 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Qualcomm® Dragonwing™ QCS8275 494.386 ms 1 - 520 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Qualcomm® Dragonwing™ QCS8550 (Proxy) 69.111 ms 12 - 582 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Qualcomm® SA8775P 126.809 ms 1 - 520 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Qualcomm® SA8650P 126.809 ms 1 - 520 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Qualcomm® SA8255P 126.809 ms 1 - 520 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Qualcomm® QCS8450 133.815 ms 3 - 514 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Qualcomm® Dragonwing™ IQ-9075 127.351 ms 12 - 25 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Qualcomm® Dragonwing™ IQ-X7181 71.038 ms 12 - 12 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Qualcomm® Dragonwing™ Q-8750 41.898 ms 0 - 520 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Qualcomm® SA7255P 494.386 ms 1 - 520 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Qualcomm® SA8295P 137.627 ms 0 - 355 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Snapdragon® 8 Elite Mobile 41.898 ms 0 - 520 MB NPU
ResNet-Mixed-Convolution QNN_DLC float Snapdragon® 8 Elite Gen 5 Mobile 31.156 ms 10 - 567 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Snapdragon® X2 Elite 26.107 ms 6 - 6 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Snapdragon® X Elite 52.097 ms 6 - 6 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Snapdragon® 8 Gen 3 Mobile 38.752 ms 6 - 718 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Snapdragon® 8 Gen 1 Mobile 76.239 ms 6 - 724 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® Dragonwing™ QCS6490 310.505 ms 6 - 13 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® Dragonwing™ QCS8275 159.737 ms 6 - 468 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® Dragonwing™ QCS8550 (Proxy) 50.591 ms 6 - 280 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® SA8775P 46.702 ms 6 - 409 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® SA8650P 46.702 ms 6 - 409 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® SA8255P 46.702 ms 6 - 409 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® QCS8450 76.239 ms 6 - 724 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-9075 47.059 ms 1 - 8 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-X7181 52.097 ms 6 - 6 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-6690 1116.926 ms 6 - 514 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-7790 89.198 ms 6 - 462 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-8750 29.804 ms 6 - 399 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® SA7255P 159.737 ms 6 - 468 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Qualcomm® SA8295P 77.607 ms 1 - 408 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Snapdragon® 8 Elite Mobile 29.804 ms 6 - 399 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Snapdragon® 8 Elite Gen 5 Mobile 21.603 ms 6 - 437 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a16 Snapdragon® 7 Gen 4 Mobile 89.198 ms 6 - 462 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Snapdragon® X2 Elite 12.492 ms 3 - 3 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Snapdragon® X Elite 26.1 ms 3 - 3 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 18.14 ms 3 - 439 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Snapdragon® 8 Gen 1 Mobile 36.031 ms 3 - 439 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS6490 113.722 ms 3 - 7 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS8275 82.757 ms 2 - 337 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 25.029 ms 3 - 6 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® SA8775P 23.452 ms 0 - 335 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® SA8650P 23.452 ms 0 - 335 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® SA8255P 23.452 ms 0 - 335 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® QCS8450 36.031 ms 3 - 439 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-9075 23.719 ms 2 - 7 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-X7181 26.1 ms 3 - 3 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-6690 494.384 ms 23 - 418 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-7790 43.758 ms 3 - 346 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-8750 15.062 ms 3 - 338 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® SA7255P 82.757 ms 2 - 337 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Qualcomm® SA8295P 42.945 ms 0 - 336 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Snapdragon® 8 Elite Mobile 15.062 ms 3 - 338 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 Mobile 10.355 ms 3 - 348 MB NPU
ResNet-Mixed-Convolution QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 43.758 ms 3 - 346 MB NPU
ResNet-Mixed-Convolution TFLITE float Snapdragon® 8 Gen 3 Mobile 1234.435 ms 0 - 829 MB NPU
ResNet-Mixed-Convolution TFLITE float Snapdragon® 8 Gen 1 Mobile 1804.079 ms 0 - 657 MB NPU
ResNet-Mixed-Convolution TFLITE float Qualcomm® Dragonwing™ QCS8275 2972.919 ms 0 - 650 MB NPU
ResNet-Mixed-Convolution TFLITE float Qualcomm® Dragonwing™ QCS8550 (Proxy) 1596.112 ms 0 - 3 MB NPU
ResNet-Mixed-Convolution TFLITE float Qualcomm® SA8775P 1565.276 ms 0 - 651 MB NPU
ResNet-Mixed-Convolution TFLITE float Qualcomm® SA8650P 1565.276 ms 0 - 651 MB NPU
ResNet-Mixed-Convolution TFLITE float Qualcomm® SA8255P 1565.276 ms 0 - 651 MB NPU
ResNet-Mixed-Convolution TFLITE float Qualcomm® QCS8450 1804.079 ms 0 - 657 MB NPU
ResNet-Mixed-Convolution TFLITE float Qualcomm® Dragonwing™ IQ-9075 1542.253 ms 0 - 43 MB NPU
ResNet-Mixed-Convolution TFLITE float Qualcomm® Dragonwing™ Q-8750 1033.943 ms 0 - 648 MB NPU
ResNet-Mixed-Convolution TFLITE float Qualcomm® SA7255P 2972.919 ms 0 - 650 MB NPU
ResNet-Mixed-Convolution TFLITE float Qualcomm® SA8295P 1819.26 ms 0 - 504 MB NPU
ResNet-Mixed-Convolution TFLITE float Snapdragon® 8 Elite Mobile 1033.943 ms 0 - 648 MB NPU
ResNet-Mixed-Convolution TFLITE float Snapdragon® 8 Elite Gen 5 Mobile 964.757 ms 0 - 657 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 2606.371 ms 1 - 1348 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Snapdragon® 8 Gen 1 Mobile 3745.494 ms 1 - 1083 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® Dragonwing™ QCS6490 16805.648 ms 3461 - 5278 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® Dragonwing™ QCS8275 6801.945 ms 3 - 1160 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 3083.448 ms 1 - 4 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® SA8775P 3424.039 ms 1 - 1123 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® SA8650P 3424.039 ms 1 - 1123 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® SA8255P 3424.039 ms 1 - 1123 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® QCS8450 3745.494 ms 1 - 1083 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® Dragonwing™ IQ-9075 3327.424 ms 0 - 38 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® Dragonwing™ Q-6690 394093.221 ms 814 - 1042 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® Dragonwing™ Q-7790 356128.919 ms 882 - 1214 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® Dragonwing™ Q-8750 2124.722 ms 1 - 818 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® SA7255P 6801.945 ms 3 - 1160 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Qualcomm® SA8295P 3748.894 ms 0 - 904 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Snapdragon® 8 Elite Mobile 2124.722 ms 1 - 818 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Snapdragon® 8 Elite Gen 5 Mobile 2819.811 ms 0 - 1317 MB NPU
ResNet-Mixed-Convolution TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 356128.919 ms 882 - 1214 MB NPU

License

  • The license for the original implementation of ResNet-Mixed-Convolution 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/ResNet-Mixed-Convolution