Hugging Face's logo

qualcomm
/
FCN-ResNet50

Image Segmentation
PyTorch
android
Model card Files
xet
 Community
FCN-ResNet50 / README.md
qaihm-bot's picture
qaihm-bot
v0.60.0
e1cdef1 verified
preview code
|
Raw
History Blame Contribute Delete
13.8 kB
metadata 
library_name: pytorch
license: other
tags:
  - android
pipeline_tag: image-segmentation

FCN-ResNet50: Optimized for Qualcomm Devices

FCN_ResNet50 is a machine learning model that can segment images from the COCO dataset. It uses ResNet50 as a backbone.

This is based on the implementation of FCN-ResNet50 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 w8a8 Universal QAIRT 2.45, ONNX Runtime 1.27.1 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 FCN-ResNet50 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 FCN-ResNet50 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

  • Input resolution: 224x224
  • Model checkpoint: COCO_WITH_VOC_LABELS_V1
  • Model size (float): 126 MB
  • Model size (w8a8): 32.2 MB
  • Number of output classes: 21
  • Number of parameters: 33.0M

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
FCN-ResNet50 ONNX float Snapdragon® X2 Elite 22.291 ms 3 - 3 MB NPU
FCN-ResNet50 ONNX float Snapdragon® X Elite 43.58 ms 63 - 63 MB NPU
FCN-ResNet50 ONNX float Snapdragon® 8 Gen 3 Mobile 32.671 ms 4 - 386 MB NPU
FCN-ResNet50 ONNX float Snapdragon® 8 Gen 1 Mobile 82.935 ms 2 - 272 MB NPU
FCN-ResNet50 ONNX float Qualcomm® Dragonwing™ IQ-8275 82.312 ms 3 - 10 MB NPU
FCN-ResNet50 ONNX float Qualcomm® Dragonwing™ QCS8550 (Proxy) 43.571 ms 0 - 79 MB NPU
FCN-ResNet50 ONNX float Qualcomm® QCS8450 82.935 ms 2 - 272 MB NPU
FCN-ResNet50 ONNX float Qualcomm® Dragonwing™ IQ-9075 74.416 ms 3 - 9 MB NPU
FCN-ResNet50 ONNX float Qualcomm® Dragonwing™ IQ-X7181 43.58 ms 63 - 63 MB NPU
FCN-ResNet50 ONNX float Qualcomm® Dragonwing™ Q-8750 26.688 ms 2 - 311 MB NPU
FCN-ResNet50 ONNX float Snapdragon® 8 Elite Mobile 26.688 ms 2 - 311 MB NPU
FCN-ResNet50 ONNX float Snapdragon® 8 Elite Gen 5 Mobile 21.051 ms 1 - 330 MB NPU
FCN-ResNet50 ONNX w8a8 Snapdragon® X2 Elite 7.247 ms 2 - 2 MB NPU
FCN-ResNet50 ONNX w8a8 Snapdragon® X Elite 13.798 ms 32 - 32 MB NPU
FCN-ResNet50 ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 10.023 ms 1 - 280 MB NPU
FCN-ResNet50 ONNX w8a8 Snapdragon® 8 Gen 1 Mobile 20.975 ms 0 - 272 MB NPU
FCN-ResNet50 ONNX w8a8 Qualcomm® Dragonwing™ QCS6490 122.126 ms 1 - 4 MB NPU
FCN-ResNet50 ONNX w8a8 Qualcomm® Dragonwing™ IQ-8275 14.373 ms 1 - 5 MB NPU
FCN-ResNet50 ONNX w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 13.596 ms 0 - 41 MB NPU
FCN-ResNet50 ONNX w8a8 Qualcomm® QCS8450 20.975 ms 0 - 272 MB NPU
FCN-ResNet50 ONNX w8a8 Qualcomm® Dragonwing™ IQ-9075 14.214 ms 1 - 4 MB NPU
FCN-ResNet50 ONNX w8a8 Qualcomm® Dragonwing™ IQ-X7181 13.798 ms 32 - 32 MB NPU
FCN-ResNet50 ONNX w8a8 Qualcomm® Dragonwing™ Q-6690 433.69 ms 1 - 384 MB NPU
FCN-ResNet50 ONNX w8a8 Qualcomm® Dragonwing™ Q-7790 25.627 ms 1 - 283 MB NPU
FCN-ResNet50 ONNX w8a8 Qualcomm® Dragonwing™ Q-8750 8.749 ms 1 - 206 MB NPU
FCN-ResNet50 ONNX w8a8 Snapdragon® 8 Elite Mobile 8.749 ms 1 - 206 MB NPU
FCN-ResNet50 ONNX w8a8 Snapdragon® 8 Elite Gen 5 Mobile 7.46 ms 0 - 249 MB NPU
FCN-ResNet50 ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 25.627 ms 1 - 283 MB NPU
FCN-ResNet50 QNN_DLC float Snapdragon® X2 Elite 22.874 ms 3 - 3 MB NPU
FCN-ResNet50 QNN_DLC float Snapdragon® X Elite 44.848 ms 3 - 3 MB NPU
FCN-ResNet50 QNN_DLC float Snapdragon® 8 Gen 3 Mobile 33.272 ms 3 - 382 MB NPU
FCN-ResNet50 QNN_DLC float Snapdragon® 8 Gen 1 Mobile 83.722 ms 3 - 272 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 82.303 ms 3 - 9 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 273.699 ms 0 - 310 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® Dragonwing™ QCS8550 (Proxy) 44.571 ms 3 - 6 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® SA8775P 72.879 ms 1 - 310 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® SA8650P 72.879 ms 1 - 310 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® SA8255P 72.879 ms 1 - 310 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® QCS8450 83.722 ms 3 - 272 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® Dragonwing™ IQ-9075 73.114 ms 3 - 8 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® Dragonwing™ IQ-X7181 44.848 ms 3 - 3 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® Dragonwing™ Q-8750 26.104 ms 0 - 320 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® SA7255P 273.699 ms 0 - 310 MB NPU
FCN-ResNet50 QNN_DLC float Qualcomm® SA8295P 77.808 ms 1 - 217 MB NPU
FCN-ResNet50 QNN_DLC float Snapdragon® 8 Elite Mobile 26.104 ms 0 - 320 MB NPU
FCN-ResNet50 QNN_DLC float Snapdragon® 8 Elite Gen 5 Mobile 21.465 ms 3 - 337 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Snapdragon® X2 Elite 7.872 ms 1 - 1 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Snapdragon® X Elite 15.197 ms 1 - 1 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 11.219 ms 1 - 263 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Snapdragon® 8 Gen 1 Mobile 24.486 ms 1 - 266 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS6490 122.932 ms 1 - 4 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-8275 15.832 ms 1 - 4 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-8275 39.492 ms 1 - 206 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 14.869 ms 1 - 2 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® SA8775P 14.736 ms 1 - 215 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® SA8650P 14.736 ms 1 - 215 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® SA8255P 14.736 ms 1 - 215 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® QCS8450 24.486 ms 1 - 266 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-9075 15.602 ms 0 - 3 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-X7181 15.197 ms 1 - 1 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-6690 410.273 ms 1 - 384 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-7790 26.035 ms 1 - 280 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-8750 8.926 ms 1 - 203 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® SA7255P 39.492 ms 1 - 206 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Qualcomm® SA8295P 21.891 ms 1 - 209 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Snapdragon® 8 Elite Mobile 8.926 ms 1 - 203 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 Mobile 7.469 ms 1 - 250 MB NPU
FCN-ResNet50 QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 26.035 ms 1 - 280 MB NPU
FCN-ResNet50 TFLITE float Snapdragon® 8 Gen 3 Mobile 33.289 ms 22 - 452 MB NPU
FCN-ResNet50 TFLITE float Snapdragon® 8 Gen 1 Mobile 83.936 ms 1 - 321 MB NPU
FCN-ResNet50 TFLITE float Qualcomm® Dragonwing™ IQ-8275 82.105 ms 0 - 72 MB NPU
FCN-ResNet50 TFLITE float Qualcomm® Dragonwing™ IQ-8275 273.585 ms 1 - 353 MB NPU
FCN-ResNet50 TFLITE float Qualcomm® Dragonwing™ QCS8550 (Proxy) 44.557 ms 0 - 87 MB NPU
FCN-ResNet50 TFLITE float Qualcomm® SA8775P 72.899 ms 0 - 354 MB NPU
FCN-ResNet50 TFLITE float Qualcomm® SA8650P 72.899 ms 0 - 354 MB NPU
FCN-ResNet50 TFLITE float Qualcomm® SA8255P 72.899 ms 0 - 354 MB NPU
FCN-ResNet50 TFLITE float Qualcomm® QCS8450 83.936 ms 1 - 321 MB NPU
FCN-ResNet50 TFLITE float Qualcomm® Dragonwing™ IQ-9075 73.02 ms 0 - 71 MB NPU
FCN-ResNet50 TFLITE float Qualcomm® Dragonwing™ Q-8750 27.1 ms 0 - 350 MB NPU
FCN-ResNet50 TFLITE float Qualcomm® SA7255P 273.585 ms 1 - 353 MB NPU
FCN-ResNet50 TFLITE float Qualcomm® SA8295P 77.758 ms 0 - 266 MB NPU
FCN-ResNet50 TFLITE float Snapdragon® 8 Elite Mobile 27.1 ms 0 - 350 MB NPU
FCN-ResNet50 TFLITE float Snapdragon® 8 Elite Gen 5 Mobile 20.929 ms 0 - 372 MB NPU
FCN-ResNet50 TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 10.987 ms 0 - 266 MB NPU
FCN-ResNet50 TFLITE w8a8 Snapdragon® 8 Gen 1 Mobile 23.148 ms 0 - 267 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® Dragonwing™ QCS6490 124.314 ms 0 - 39 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® Dragonwing™ IQ-8275 15.175 ms 0 - 37 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® Dragonwing™ IQ-8275 38.226 ms 0 - 206 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 14.272 ms 0 - 2 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® SA8775P 14.698 ms 0 - 207 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® SA8650P 14.698 ms 0 - 207 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® SA8255P 14.698 ms 0 - 207 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® QCS8450 23.148 ms 0 - 267 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® Dragonwing™ IQ-9075 15.067 ms 0 - 36 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® Dragonwing™ Q-6690 428.846 ms 0 - 389 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® Dragonwing™ Q-7790 27.887 ms 0 - 280 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® Dragonwing™ Q-8750 8.577 ms 0 - 198 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® SA7255P 38.226 ms 0 - 206 MB NPU
FCN-ResNet50 TFLITE w8a8 Qualcomm® SA8295P 21.306 ms 0 - 211 MB NPU
FCN-ResNet50 TFLITE w8a8 Snapdragon® 8 Elite Mobile 8.577 ms 0 - 198 MB NPU
FCN-ResNet50 TFLITE w8a8 Snapdragon® 8 Elite Gen 5 Mobile 7.413 ms 0 - 246 MB NPU
FCN-ResNet50 TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 27.887 ms 0 - 280 MB NPU

License

  • The license for the original implementation of FCN-ResNet50 can be found here.

References

Community